I am trying to construct the induced representations of the dihedral group $G=D_p$, $|D_p|=2p$, if I take the subgroup $H=\langle r \rangle \cong C_p$, which is generated by the rotations. I have problems as I try to understand the general way of the construction, so I try it with this example. It's not a homework!

If $W$ is a $\mathbb{Q}H$-module, then the induced representation is defined as $\mathrm{ind}^G_H (W):=W \otimes_{\mathbb{Q}H} \mathbb{Q}G$. In this example $W$ has to be $\mathbb{Q}$ or $\mathbb{Q}(\zeta_p)$, because these are the simple $\mathbb{Q}H$-modules. For $k=\dim(W)$ and $l=[G:H]$ I can choose a $\mathbb{Q}$-basis $w_1,\cdots , w_k$ of $W$ and a $\mathbb{Q}H$-basis $g_1,\cdots , g_l $ of $\mathbb{Q}G$. Then $$w_1 \otimes g_1,\cdots ,w_k \otimes g_1,\cdots w_k \otimes ,g_l $$ has to be a $\mathbb{Q}$-base of $\mathrm{ind}^G_H(W)$ and $\dim(\mathrm{ind}^G_H(W))=kl$.

So if I first take $W=\mathbb{Q}(\zeta_p)$, what would the induced representation be? I think, that the dihedral group should have two $1$-dim and one $(p-1)$-dim representations. How can I get the $(p-1)$-dim one?

Somehow I cannot write any comments:

Thanks for your help, but I have to work with representations over $\mathbb{Q}$ and later over $\mathbb{R}$. To get representations over $\mathbb{C}$, I could avoid to use the induced representation and for many groups it was not really difficult to get the irreducible representations without it.

Of course I know Maschke's theorem, but it gives me only the existence of irreducible representations, if e.g. $char(G)=0$, but not how to construct them. I hope anyone has an idea, how to work with $ind(W)$ in this expamle?

  • $\begingroup$ I'm curious: what introductory text works over $\mathbf Q$ and not $\mathbf C$? $\endgroup$ Apr 18, 2012 at 20:16
  • $\begingroup$ I don't have an "introductory text", just a general definition of the induced representation over a field $\mathbb{K}$. The representations over $\mathbb{C}$ you can usually get without it, so I try to understand it over $\mathbb{Q}$, or are there many differences? $\endgroup$
    – bounty
    Apr 18, 2012 at 20:26
  • $\begingroup$ @bounty: This theory is best over algebraically closed fields. So while you don't need $\mathbb{C}$, you do need to at least work in $\overline{\mathbb{Q}}$ in order to have basic results like Maschke's Theorem (and in fact, your representations will live in this field). $\endgroup$ Apr 18, 2012 at 21:48
  • 1
    $\begingroup$ @BrettFrankel, Maschke's Theorem holds for a field of characteristic zero without any restrictions... $\endgroup$ Apr 18, 2012 at 23:37
  • 1
    $\begingroup$ @bounty Of course it does, silly me. I was thinking Schur's Lemma. $\endgroup$ Apr 18, 2012 at 23:39

1 Answer 1


The irreducible $(p-1)$-dimensional $\mathbb{Q}G$-module is not induced from a rational representation of $C_p$. But twice that is. One way of seeing this is to work over $\mathbb{C}$. Over $\mathbb{C}$, the $(p-1)$-dimensional guy decomposes as a sum of $\frac{p-1}{2}$ irreducible 2-dimensional representations. Each is induced from a one-dimensional representation of $C_p=H$. However, if $\chi$ is a non-trivial one-dimensional character of $C_p$, then $\mathrm{Ind}_{G/H}\chi = \mathrm{Ind}_{G/H}\bar{\chi}$. That follows from Clifford theory: $H$ is normal in $G$ and $\chi$ and $\bar{\chi}$ lie in one orbit under the action of $G$ on the irreducible characters of $H$.

Thus, if you induce the sum of all non-trivial one-dimensional representations of $H$ to $G$, you hit each irreducible 2-dimensional complex representation of $G$ twice.

Another way of obtaining the $(p-1)$-dimensional guy through induction is by inducing the trivial representation of $C_2\leq G$. This induction will be $p$-dimensional and will have the trivial $G$-representation as a direct summand (this uses Maschke's theorem, which of course works just fine over $\mathbb{Q}$). The remaining $(p-1)$-dimensional chunk is irreducible over $\mathbb{Q}$, and is the representation you are looking for.


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.