3

In class we proved the following theorem:

Let $σ:G\rightarrow GL(n,\mathbb{C})$ be a representation of $G$ and let $χ$ be the corresponding character.
Then $\forall g\in G$ we have: If $0<\frac{|χ(g)|}{χ(1)}<1$ then $\frac{χ(g)}{χ(1)}\notin\overline{\mathbb{Z}}$

The proof is pretty straight forward and goes as follows.

First fix $g\in G$ and set $a=\frac{χ(g)}{χ(1)}=\frac{χ(g)}{n}$.
We know that $χ(g)=\sum_{i=1}^{n}ω_i$, where each $ω_i$ is a root of unity.
We'll show that if $\frac{|χ(g)|}{χ(1)}<1$ and $\frac{χ(g)}{χ(1)}\in\overline{\mathbb{Z}}$ then $χ(g)=0$.

Let $p(x)=x^{k}+a_{k-1}x^{k-1}+\dots+a_{1}x+a_{0}$, $a_{i}\in\mathbb{Z}$ be the minimal polynomial of $a$.
Since $a=\frac{ω_{1}+\dots+ω_{n}}{n}$, then every other root of $p(x)$ is also of the form $\frac{ω_{1}'+\dots+ω_{n}'}{n}$ and hence we get that $|a_0|<1$, since $a_{0}=$product of all roots.

(Note: Actually to be exact, the proof we were given doesn't say "that since $a$ is of that form, then the rest of the roots must also be like that".
It rather says "Sicne every other root of $p(x)$ is also of that form, then...", but I assume it's the same thing.)

But $a_{0}\in\mathbb{Z}$ so $a_{0}=0$, and since $p(x)$ is irreducible we must have that $p(x)=x$ which implies that $a=0\Rightarrow χ(g)=0$.

Now I understand every part of it, apart from the implication that since $a$ is the sum of $n$ roots of unity divided by $n$, then the same must hold for the other roots.
How do we deduce such a claim?

Thank you in advance.

Fotis
  • 953

1 Answers1

4

This has something to do with Galois theory. Since the minimal polynomial is irreducible, its roots are permuted transitively by the Galois group of the splitting field. But the Galois group sends roots of unity to roots of unity.

Brauer Suzuki
  • 4,425
  • 1
  • 9
  • 28
  • Yes, by $\overline{\mathbb{Z}}$ I mean the set of algebraic integers.

    A much simpler proof indeed, nice, I like this one a lot more haha. But still, could you explain to me the original question, namely why all the roots of the minimal polynomial are of the form (sum of roots of unity)/n?

    After seeing your answer its probably connected with the conjugate pairs of roots that appear but I'm still not quite sure on how to prove the claim.

    Other than that, thank you for the alternate (much easier to understand) proof

    – Fotis May 06 '23 at 17:44
  • 2
    This has something to do with Galois theory. Since the minimal polynomial is irreducible, its roots are permuted transitively by the Galois group of the splitting field. But the Galois group sends roots of unity to roots of unity. – Brauer Suzuki May 06 '23 at 18:09
  • 1
    ooh yeah that's right, Galois Theory didn't even cross my mind, now it makes so much sense. Perfect thank you very much for both of your answers! – Fotis May 06 '23 at 18:23
  • Do you want to write that comment as an answer as well, since technically this answers the original question? Or should I add it and mentions you. Or maybe its not even necessary, I'm not sure – Fotis May 06 '23 at 18:25
  • well, my simpler proof was wrong! The product is a real number, but not necessarily a rational number. I now turned my comment into the answer. – Brauer Suzuki May 07 '23 at 04:44