3
$\begingroup$

Let $G$ be a finite group with normal subgroup $N$. Let $\chi$ be an irreducible complex character of $G$. Then Clifford's Theorem says that $\mathrm{res}^{G}_{N}\chi = e(\eta_1 + \cdots + \eta_r)$ where $e$ is a positive integer and each $\eta_i$ is a complex irreducible character of $N$ such that the $\eta_i$'s are the distinct conjugates of $\eta := \eta_1$ (i.e. for each $i$ there exists $g_i \in G$ such that $\eta_i(n)=\eta(g_i n g_i^{-1})$ for all $n \in N$.) Let $\mathbb{Q}(\eta) = \mathbb{Q}(\eta(n) : n \in N)$ be the character field of $\eta$ and similarly for $\mathbb{Q}(\mathrm{res}^{G}_{N} \chi)$. Since $\mathbb{Q}(\eta_i)=\mathbb{Q}(\eta)$ for each $i$, we have $K:=\mathbb{Q}(\mathrm{res}^{G}_{N} \chi) \subseteq \mathbb{Q}(\eta)$.

Question: Is $e$ somehow related to $[\mathbb{Q}(\eta):K]$? For example, do we have $e \geq [\mathbb{Q}(\eta):K]$? Otherwise, can you tell me anything about the relation between $K$ and $\mathbb{Q}(\eta)$? e.g. do we somehow get $\mathbb{Q}(\eta)$ from $K$ by adjoining certain roots of unity? If so, which ones (are they related to say $[G:N]$)? (I am actually interested in the case where we replace $\mathbb{Q}$ by the $p$-adics $\mathbb{Q}_p$ throughout and $[G:N]$ is a power of $p$.)

$\endgroup$

3 Answers 3

6
$\begingroup$

I don't think there's any clear cut relationship. In particular, the inequality $e \ge [\mathbb Q(\eta) : K]$ needn't hold. For example, take $G=S_5$, $N=A_5$ and let $\chi$ be the unique irreducible $S_5$-character of degree $6$. Then we have $\mathrm{res}^{G}_{N} \chi = \eta + \bar{\eta}$, where $\eta$ and $\bar{\eta}$ are the degree $3$ irreducible characters of $A_5$. So here $e=1$ and $\mathbb Q(\eta) = \mathbb Q(\sqrt 5)$. On the other hand, $\chi$ is integer valued so $K=\mathbb Q$. Thus $e < [\mathbb Q(\eta) : K]$ in this case.

$\endgroup$
1
  • $\begingroup$ Thanks for the counterexample! As you say, it looks like the relationship is not clear in general. $\endgroup$ Mar 18, 2013 at 13:05
7
$\begingroup$

The coefficient $e$ is always $1$ if $G/N$ is cyclic, but $|Q(\eta):Q(\chi_N)|$ can be arbitrarily large. For example, let $N$ be cyclic of prime order $p$, and let $G$ be the semidirect product of a cyclic group of order $p-1$ acting faithfully on $N$. Then $G$ has a rational valued character $\chi$ of degree $p-1$ and the field $Q(\eta)$ is the cyclotomic field of $p$th roots of unity. Then $|Q(\eta):Q(\chi_N)| = |Q(\eta):Q| = p-1$.

It is also true that $e$ can be much larger than $|Q(\eta):Q(\chi_N)|$. For example, take $G$ to be extraspecial of order $2^{2n+1}$ and $N = Z(G)$. Here $e = 2^n$ and both $Q(\eta)$ and $Q(\chi_N)$ are just the rationals $Q$.

$\endgroup$
1
$\begingroup$

A general technique for handling such problems is the following: Let $T=G_{\eta}$ be the inertia subgroup of $\eta$ and $\hat{T}$ the inertia group of its Galois orbit, that is, $\newcommand{\QQ}{\mathbb{Q}} \DeclareMathOperator{\Gal}{Gal}$ $$ T = \{g\in G \mid \eta^g = \eta \} \quad \text{and}\quad \hat{T} = \{ g\in G \mid \exists \alpha_g \in \Gal(\QQ(\eta)/\QQ ) \colon \eta^{g\alpha_g}=\eta \}.$$ By Clifford theory, there is a unique character $\psi$ of $T$ such that $\chi=\psi^G$ and $\psi$ lies over $\eta$. Set $\hat{\psi}:= \psi^{\hat{T}}$. Then the following hold:

  1. $\QQ(\chi)=\QQ(\hat{\psi})$ and $\QQ(\psi) = \QQ(\chi, \eta)$.

  2. The map $\hat{T}\ni t \mapsto \alpha_t\in \Gal( \QQ(\eta)/ \QQ( \chi_N) )$ is a group homomorphism with kernel $T$ (in general not surjective). In particular, $T$ is normal in $\hat{T}$ with abelian factor group.

  3. The subgroup $\Gal( \QQ(\eta) / \QQ(\eta) \cap \QQ(\chi) )$ is contained in the image of $\hat{T}\ni t \mapsto \alpha_t$. (Notice that this Galois group is isomorphic to $\Gal( \QQ(\psi) / \QQ(\chi) )$ by 1.)

This can be shown by elementary Clifford and Galois theory ( Theorem 1 in: Udo Riese and Peter Schmid, MR 1388863 Schur indices and Schur groups. II, J. Algebra 182 (1996), no. 1, 183--200. )

It has been pointed out in the other answers that the relation between the $e$ and the fields is essentially arbitrary. The extension $\QQ(\eta)/\QQ(\chi_N)$ is also arbitrary, except that both fields must be contained in some cyclotomic extension of $\QQ$: Suppose that $\QQ \subseteq \mathbb{F}\subseteq \mathbb{E} \subseteq \QQ(\epsilon)$, where $\epsilon$ is a primitive $n$-th root of unity. Let $G$ be the semidirect product of $\Gal( \QQ(\epsilon) / \mathbb{F} )$ with $C_n =\langle \epsilon\rangle$, and $N$ the semidirect product of $\Gal( \QQ(\epsilon) /\mathbb{E} )$ with $C_n$. Then every faithful linear character of $C_n$ induces to an irreducible character $\chi$ of $G$ with $\QQ(\chi_N)=\mathbb{F}=\QQ(\chi)$ and $\QQ(\eta) =\mathbb{E}$.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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