Definition : Let $\sim$ be the equivalence relation on inclusions of finite groups, generated by :
$(H \subset G) \sim (\phi(H) \subset \phi(G))$, with $ \phi: G \to L$ a group morphism, and $ker(\phi) \subset H$.

Question : $(\langle (1234) \rangle \subset S_4 ) \sim (\langle (12),(34) \rangle \subset S_4 )$ ?

Warning: Of course, there is no $\phi \in Aut(S_4)$ with $\phi(\langle (1234) \rangle) = \langle (12),(34) \rangle$, because $\langle (1234) \rangle \simeq \mathbb{Z}_4 \not\simeq \mathbb{Z}^2_2 \simeq \langle (12),(34) \rangle$, that why $\sim$ is the equivalence relation generated by the relation above, not just the relation alone.

Remark : the non-trivial normal subgroups of $S_4$ are $A_4$ and $\langle (12)(34), (13)(24) \rangle$ (see here).

Motivation : These inclusions are famous in the group-subgroup subfactor theory (see here p47).

  • 1
    $\begingroup$ No. If $H \subset G$ is equivalent to $X \subset Y$ then the permutation action of $G$ on the cosets of $H$ is equivalent to the permutation action of $Y$ on the cosets of $X$. However, those two groups of degree 6 are not permutation isomorphic. $\endgroup$ – Jack Schmidt Feb 15 '14 at 22:16
  • $\begingroup$ @JackSchmidt : I have developed your comment as an answer, I hope it's what you had in mind. $\endgroup$ – Sebastien Palcoux Feb 16 '14 at 15:41

I'm not an expert in finite groups theory, so I will try to develop the comment of Jack Schmidt :

Let $(H \subset G)$ and $\phi:G \to L$ with $ker(\phi) \subset H$.
Let $\Omega=G/H$ and $\Omega' = \phi(G) / \phi(H)$.

Let the map $\delta : \Omega \to \Omega'$ : $\delta(gH) = \phi(g)\phi(H)$
$\delta$ is well-defined because if $g_1^{-1}g_2 \in H$ then $\phi(g_1)^{-1}\phi(g_2) = \phi(g_1^{-1}g_2) \in \phi(H)$
$\delta$ is then obviously surjective, and it's injective because if $\phi(g_1)^{-1}\phi(g_2) \in \phi(H)$ then $g_1^{-1}g_2 \in H$, because for $h \in H$, $\phi^{-1} (\{\phi (h) \}) = h.ker(\phi) \subset H$.

Let $\pi: G \to S_{\Omega} : \pi(g_1)(g_2H) = g_1g_2H$, then $\pi(G) \simeq G/ker(\pi) $
$ker(\pi) = \{ g_1 \in G: g_1g_2H = g_2H, \forall g_2 \in G \} = \bigcap_{g_2 \in G} g_2^{-1}Hg_2 = core_{G}(H) \subset H$ is the normal core of $H$ in $G$ i.e the largest normal subgroup of $G$ contained in $H$.
Idem, let $\pi': \phi(G) \to S_{\Omega'} $, then $\pi'(\phi(G)) \simeq \phi(G)/ker(\pi') $

Now $\pi(G)$ and $\pi'(\phi(G))$ are permutation isomorphic because $\delta$ is a bijection and obviously $\delta(\pi(g_1)(g_2H)) = \pi'(\phi(g_1))(\phi(g_2)\phi(H))$; in particular $\pi(G) \simeq \pi'(\phi(G))$.

Conclusion, the equivalence relation generated by: $$(H \subset G) \sim (\phi(H) \subset \phi(G)) \text{ with } \phi: G \to L \text{ morphism and } ker(\phi) \subset H$$ implies this equivalence relation $\sim'$ as above (equivalence of permutation actions on the cosets).

Optional question : What's about the converse ?

Anyway, about my example :
If $(\langle (1234) \rangle \subset S_4 ) \sim (\langle (12),(34) \rangle \subset S_4 )$, then $(\langle (1234) \rangle \subset S_4 ) \sim' (\langle (12),(34) \rangle \subset S_4 )$ and so $\langle (1234) \rangle \simeq \langle (12),(34) \rangle$, because the normal cores are trival, contradiction.

The normal cores are trival because there is not non-trivial normal subgroup of $S_4$ contained in $\langle (1234) \rangle $ or $ \langle (12),(34) \rangle$, thanks to the classification of the normal subgroups of $S_4$ given above.

  • 1
    $\begingroup$ This seems fine. I think the equivalence relation looks like $(H \subset G) \to (H/H_G \subset G/H_G) \cong (X/X_Y \subset Y/X_Y) \leftarrow (X \subset Y)$ where $K_L$ is the largest normal subgroup of $L$ contained in $K$. In other words, every inclusion is equivalent to a core-free inclusion, and two inclusions are equivalent iff there core-free inclusions are permutation isomorphic. $\endgroup$ – Jack Schmidt Feb 16 '14 at 15:46

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.