I am working through the following representation theory notes: Notes

I am having some trouble understanding why the Corollary to Theorem 15 is true.

Theorem 15 Suppose $G$ is a finite group with irreducible characters $\chi_{1},\dots \chi_{k}$ over $\mathbb{C}$ If $\chi$ is any character, expressible as a sum of irreducible characters by $\chi = \Sigma m_{i} \chi_{i} $ then:

  1. For each i , $ m_{i}=\langle \chi_{i} | \chi_{i} \rangle $
  2. $\langle \chi | \chi \rangle = \Sigma m_{i}^{2} $

Corollary A character $\chi$ is irreducible if and only if $\langle \chi | \chi \rangle=1$

I understand that if $\chi$ is irreducible then it must be one of $\chi_{1},\dots \chi_{k}$ and then $\langle \chi | \chi \rangle=1$

I am having trouble understanding why the converse must be true. For example if I had $\chi =\frac{3}{5}\chi_{1}+\frac{4}{5}\chi_{2} $ then $\langle \chi | \chi \rangle = \Sigma m_{i}^{2}=1 $ but this is reducible.

I would appreciate some clarification on what is going on and a 'proof' of the corollary. Thank you.


You are not allowed to have rational numbers in the linear combination of the characters. The $m_i$'s are natural numbers. Theorem 10 in your notes says that

Every character is a sum of irreducible characters. 

So you can't have half of a character.

So then, if $\chi = m_1\chi_1 + \dots + m_k\chi_k$, you see that $\chi$ is irreducible if and only if exactly one $m_i =1$ and the rest zero.

  • $\begingroup$ Why the downvote? $\endgroup$ – Thomas Sep 9 '15 at 17:01
  • $\begingroup$ @Ravi: Nice catch! lol $\endgroup$ – Thomas Jul 8 '17 at 13:01

Any arbitrary (non-virtual) character is a linear sum $\chi = \sum m_i \chi_i$ with the $\chi_i$ irreducible characters and the $m_i$ positive integers. In terms of the underlying representations, any (complex) representation $\rho$ is the direct sum $\rho = \rho_1^{\oplus n_1} \oplus \cdots \oplus \rho_k^{\oplus n_k}$ with each $\rho_i$ an irrep and the $n_i$ positive integers; taking traces gives $\chi_{\rho} = n_1\chi_{\rho_1} + \cdots + n_k \chi_{\rho_k}$ for the corresponding characters. Given that, any character $\chi = \sum m_i \chi_i$ with $\langle{\chi, \chi}\rangle = 1$ must have $\sum m_i^2 = 1$ and thus exactly one nonzero $m_i = 1$.

A character is specifically the trace of a representation; not every class function on the group $G$ is a character. This does become true after tensoring with $\mathbb{C}$ (that is, the set of irreducible characters form a basis for the vector space of class functions on $G$) but an arbitrary sum of characters is not necessarily itself a character. For example, the values of characters are algebraic integers.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.