So I've been trying to answer this exercise to much of my frustration:

Let $G$ be a finite group and $S$ a normal subgroup. Let $\rho$ be an irreducible representation of $G$ over $\mathbb{C}$. Prove that either the restriction of $\rho$ to $S$ has all its irreducible components $S$-isomorphic to each other, or there exists a proper subgroup $H$ of $G$ containing $S$ and an irreducible representation $\theta$ of $H$ such that $\rho \simeq \text{ind}_H^G(\theta)$.

Here is my progress so far:

Let $E$ be a representation space for $\rho$ and $\chi_\rho$ the character. We also have the restriction $\text{res}_S^G(\rho)$ of our representation $\rho$ to $S$. $\text{res}_S^G(E)$ is the representation space for this restriction.

We pick a simple $S$-submodule $F$ of $\text{res}_S^G(E)$ and realize that because $E$ is a simple $G$-module, \begin{equation} \text{res}_S^G(E)=\sum_i \gamma_i F \end{equation} where $\{\gamma_i\}$ is a set of left coset representatives for $G/S$. Also note that $S\trianglelefteq G$ implies that $\gamma_i F$ is an irreducible $S$-submodule of $\text{res}_S^G(E)$. If these submodules are $S$-isomorphic to each other, then we have the first case.

Now suppose otherwise. I am guessing that we are able to find a subgroup $S\subseteq H\subsetneq G$ and an irreducible character $\chi_\theta$ in the character decomposition of $\chi_{\text{res}_H^G(\rho)}$ such that $\text{ind}_H^G(\chi_\theta)$ is simple. We would then have \begin{equation} \langle\text{res}_H^G \chi_\rho,\chi_\theta\rangle=\langle\chi_{\text{res}_H^G(\rho)},\chi_\theta\rangle\geq 1. \end{equation} By Frobenius reciprocity, we would have $\langle \chi_\rho,\text{ind}_H^G(\theta) \rangle \geq 1$, which implies $\langle \chi_\rho,\text{ind}_H^G(\chi_\theta) \rangle = 1$ since both characters are irreducible. If $\theta$ is a representation corresponding to $\chi_\theta$, then $\rho\simeq \theta$.

Now it remains to find this subgroup $H$. I have a hunch that $H=\{\sigma\in G: \sigma F \simeq F\text{ as an $S$-representation space}\}$. It can be easily shown that $H\subsetneq G$ is a proper subgroup. However, I am not sure how to find such an irreducible character $\chi_\theta$ with our desired properties. Any hints are appreciated!


1 Answer 1


Your hunch on $H$ is correct. For completeness, I redo the problem.

Let $\rho: G\to \text{GL}(V)$ be the representation given in question, consider the canonical decomposition $V = \bigoplus_i V_i$ when $V$ is viewed as an $S$-module. $V_i$ is a direct sum of isomorphic irreducible representations of $S$, and every irr-rep of $S$ in this isomorphism class is contained in $V_i$.

Because $S\lhd G$, $G$ permutes $V_i$. (Proof: if $W\subset V_i$ is a simple $S$-module, then normality shows $gW$ is also a simple $S$-module, and if $W_1, W_2$ are $S$-isomorphic, then so are $gW_1, gW_2$. Summing over all $W$ gives the claim.) The action of $G$ on $\{V_i\}$ is transitive since $\rho$ is irreducible.

If there is only one $V_i$, then we are done. Otherwise, fix one $V_1$, let $$H = \{ g\in G | gV_1 = V_1\}$$ satisfies $S\subset H$, $H\neq G$. $\text{Ind}_H^G(V_1)$ is obviously $\rho$. Therefore it remains to prove $V_1$ is a simple $H$-module. If $W_0\subset V_1$ is a proper subspace invariant under $H$, then $W_0$ contains a simple $S$-module $W$. Then $$g\in H \iff gV_1 = V_i \iff gW\subset V_1$$ Since $\rho$ is irreducible, this forces $W_0 = V_1$. Completing the proof.


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.