4
$\begingroup$

Consider the following claim: a finite group $G$ is Dedekind $\iff$ for every irrep $\rho$, and every $g \in G$, $\rho(g)$ either is identity matrix or has all eigenvalues $\ne 1$.

Is this claim true?

Recall that finite Dedekind groups are abelian or $Q_8 \times \mathbb{Z}_2^t \times \mathbb{Z}_m$ for odd $m$, integer $t$, $Q_8$ the quaternion group.

Dedekind $\implies$ for every irrep $\rho$, and every $g \in G$, $\rho(g)$ either is identity matrix or has all eigenvalues $\ne 1$ holds by inspection. So the question is whether $\impliedby$ holds.

$\endgroup$
2
  • 2
    $\begingroup$ Could you remind us of the definition of "Dedekind group"? $\endgroup$
    – Yemon Choi
    Commented Nov 11 at 23:46
  • 4
    $\begingroup$ The definition you wrote does not agree with wikipedia (a finite abelian group of odd order need not be cyclic). $\endgroup$
    – Aurel
    Commented Nov 12 at 7:33

1 Answer 1

7
$\begingroup$

The condition can be restated as: for every $g$ and $\rho$, the subspace $\mathrm{Ker}(\rho(g)-1)$ is a subrepresentation.

If $G$ is Dedekind (meaning that every subgroup is normal, or equivalently that every cyclic subgroup is normal), then $\langle g\rangle$ being normal, the result follows (no need to use inspection).

Conversely, suppose that $\mathrm{Ker}(\lambda(g)-1)$ is a subrepresentation for $\lambda$ the left regular representation. This kernel is the set $U_g$ of functions that are left $\langle g\rangle$-invariant, and we readily see that for $h\in G$, $hU_g=U_{hgh^{-1}}$, and $U_g=U_{g'}$ iff $\langle g\rangle =\langle g'\rangle$. So the condition that $U_g$ is a subrepresentation means that $\langle g\rangle$ is normal. If this is true for every $g$, this means that $G$ is Dedekind.

$\endgroup$
5
  • $\begingroup$ Clifford's Theorem is relevant here and might be illuminating. $\endgroup$ Commented Nov 13 at 0:08
  • $\begingroup$ @GeoffRobinson Pointer? The Wikipedia reference for Clifford's theorem seems unrelated. [But well "illuminating": does this straightforward proof really seem obscure?] $\endgroup$
    – YCor
    Commented Nov 13 at 0:39
  • $\begingroup$ No, your proof is fine and certainly not obscure, I intended no criticism of it, rather the opposite, since I updated it. To me Clifford's Theorem provides a natural explanation for the fact that that a normal subgroup with any fixed point in an irreducible representation acts trivially, but it is true that CT might be considered unnecessary machinery to prove that particular fact. I have never looked at the Wikid $\endgroup$ Commented Nov 13 at 1:41
  • $\begingroup$ ....Wikipedia page for CT, but for me CT is an indispensable tool in representatio theory- that may be a matter of taste. -I meant up voted it that was an auto text error. $\endgroup$ Commented Nov 13 at 1:50
  • $\begingroup$ @GeoffRobinson I finally found the link Clifford theory. The theorem assumes $N$ normal so indeed reproves the most trivial implication (with a far-reaching generalization). $\endgroup$
    – YCor
    Commented Nov 13 at 8:26

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .