3
$\begingroup$

Could you tell me how to show that $(123)$ and $(132)$ are not in the same conjugacy class in $A_4$?

I know that all 3-cycles can't be in the same class, because the order of each class must divide $|A_4| = 12$ and there are eight 3-cycles in $A_4$.

If $\sigma$ is a 3-cycle fixing $n$ and $τ \in A_4$ then $\tau \circ \sigma \circ \tau^{-1}$ is a 3-cycle fixing $τ(n)$. (And there are 3 such permutations). If we conjugate $(123)$ by $(132)$ we will get $(123)$ and vice versa. So it would appear that it only makes sense to conjugate $(123)$ by something with a $4$. If we conjugated $(123)$ by $(124)$, we would need $(124)(123)(124)^{-1}$ to send 4 to 4, but it doesn't. It turns out that such $\tau$ cannot have anything "common" with $(123)$ - what I mean by this is that there can't be $(12...), (23...), (31...)$ in $\tau$. So the only cycles that remain are $(134), (142), (243)$ - and they are indeed ion the same class as $(123)$.

As you see, this isn't the best way to solve this problem.

I also know that if $g \in A_n$ commutes with an odd permutation, then all permutations with the same cycle type as $g$ are in one conjugacy class, but I have no idea how to show that in $A_4$ no odd permutation commutes with $(123)$.

Could you make it more mature?

Thank you.

$\endgroup$
  • 1
    $\begingroup$ You can fairly easily see what a permutation must look like in order to conjugate $(123)$ to $(132)$, and then check that none of the possibilities are even permutations. $\endgroup$ – Tobias Kildetoft Oct 6 '13 at 10:11
  • $\begingroup$ Well. I know that such a permutation must stabilize one of $1,2,3$ and of course $4$, because $\sigma (123) \sigma^{-1} = (\sigma(1) \sigma (2) \sigma (3)) = (132)$. Do you think it's enough? $\endgroup$ – Bilbo Oct 6 '13 at 10:18
  • 1
    $\begingroup$ Since once you note that it must fix $2$ elements (and the identity does not work), you can conclude that it must be a transposition, yes, that is enough. $\endgroup$ – Tobias Kildetoft Oct 6 '13 at 10:20
  • $\begingroup$ Ok, thank you a lot :) $\endgroup$ – Bilbo Oct 6 '13 at 10:21
3
$\begingroup$

One way to see this is to work in $S_4$ instead. There $(1\,2\,3)$ and $(1\,3\,2)$ are conjugates, of course, since $(2\,3)^{-1}(1\,2\,3)(2\,3)=(1\,3\,2)$.

If we have a $\tau$ such that $\tau^{-1}(1\,2\,3)\tau=(1\,3\,2)$, it must factor as $\tau=\rho(2\,3)$ for some $\rho$, and we have $$(2\,3)^{-1}\rho^{-1}(1\,2\,3)\rho(2\,3)=(1\,3\,2)=(2\,3)^{-1}(1\,2\,3)(2\,3)$$ Canceling the $(2\,3)$s we find $\rho^{-1}(1\,2\,3)\rho=(1\,2\,3)$, and it's then easy to see that $\rho(1)$ determines $\rho$ and there are only three possibilities, which are all even permutations. So $\tau$ must be odd and so not in $A_4$.


This may or may not be easier to follow than Tobias' suggestion to brute-force the possible $\tau$s directly. Personally I think it's easier to think about possible stabilizers of a cycle than to keep things straight while trying to conjugate one cycle into another.

$\endgroup$
  • $\begingroup$ I've already accepted your answer, but now I see that I don't quite understand why $\tau = \rho (23)$. Could you explain that? Does it simply follow from from $(2\,3)^{-1}(1\,2\,3)(2\,3)=(1\,3\,2)$? $\endgroup$ – Bilbo Oct 6 '13 at 10:26
  • $\begingroup$ @Bilbo: In a group, everything factors through everything else. Whenever you have a $\tau$ you can just use $\tau(2\,3)^{-1}$ as $\rho$. $\endgroup$ – Henning Makholm Oct 6 '13 at 10:27
  • $\begingroup$ All right. Thank you. $\endgroup$ – Bilbo Oct 6 '13 at 10:29
1
$\begingroup$

For an easy argument, you can use that the size of the conjugacy class $C_G(g)$ of $g\in G$ is $|G|/|Z_G(g)|$ where $Z_G(g)=\{\,h\in G\mid gh=hg\,\}$ is the centraliser of$~g$ (this is a special case of the orbit-stabiliser theorem). The centraliser contains at least the powers of $g=(123)$, which here gives $|Z_{A_4}(g)|\geq3$ and therefore $|C(g)|\leq12/3=4$. So as soon as we have $4$ conjugates of $g$ there cannot be more of them; one easily sees that $\{(123),(134),(142),(243)\}\subseteq C_{A_4}(g)$, and hence this must be all.

$\endgroup$
  • $\begingroup$ That's a very nice argument. Thank you. $\endgroup$ – Bilbo Oct 6 '13 at 14:27

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.