5
$\begingroup$

Suppose $G=S_n$ is the permutation group in $n$ letters and $7\leq n\leq 10$. Also consider the subgroups $H_1,H_2$ such that $H_1$ is isomorphic with $S_k$ and $H_2$ is isomorphic with $S_{n-k}$ and every element of $H_1$ commutes with every element of $H_2$. Is it true that $H_1,H_2$ are the stabilizers of $n-k$ and $k$ letters respectively?

$\endgroup$
11
  • 1
    $\begingroup$ For $k=1,2,3,\lfloor n/2\rfloor$, you should consider all possible actions of $S_{n-k}$ (or $S_k$, if easier) on $n$ elements, compute their centralizers and check what's going on. What have you tried so far? $\endgroup$
    – YCor
    Sep 2, 2016 at 9:17
  • 2
    $\begingroup$ $S_5$ has a transitive action in $S_6$, and the centralizer is $S_1$ (i.e. the trivial group). So there's a counter-example. There might be others for small $n$. But I'd expect the statement is true for $n$ large enough -- use the fact that the centralizer of a transitive group is semi-regular. $\endgroup$
    – Nick Gill
    Sep 2, 2016 at 9:25
  • 3
    $\begingroup$ By the way, I don't understand why there is a vote to close. The question seems perfectly reasonable to me. $\endgroup$
    – Nick Gill
    Sep 2, 2016 at 9:27
  • 2
    $\begingroup$ @YCor, Fair enough. I think the question is OK here. But MSE would be fine too. $\endgroup$
    – Nick Gill
    Sep 2, 2016 at 9:36
  • 1
    $\begingroup$ A similar idea to Nick's gives a counterexample with $k=2$ or $k=3$. $S_6$ has an outer automorphism (this sends the stabilizer of a point to the transitive $S_5$ Nick mentioned). So consider stabilizers of $6-k$ and $k$ letters in $S_6$ (for $k=2$ and $k=3$) and then take their images under the outer automorphism. $\endgroup$ Sep 2, 2016 at 12:14

1 Answer 1

1
$\begingroup$

Pick a prime $p$ in the range $\frac{n}{2}<p\leq n-k$. Then $S_{n-k}$ contains $(p-1)!\binom{n-k}{p}$ elements of order $p$, and elements of order $p$ in $S_n$ are $p$-cycles. There are at most $(p-1)!$ cycles of length $p$ in $H_2$, which move the same $p$ points, hence $H_1$ fixes the union of $\geq\binom{n-k}{p}$ sets of size $p$ pointwise, and this union has $\geq n-k$ elements. Hence $H_1$ is the stabilizer of a set of $\geq n-k$ points, and $>$ is clearly impossible.

If there is no prime in this range, things get a bit more complicated. Suppose $p<q$ are primes with $p+q\leq n-k$. Then an element of order $pq$ in $H_2$ consists of $\alpha$ cycles of length $p$ and $\beta$ cycles of length $q$. If $\alpha=\beta=1$, you can argue as above. Otherwise the centralizer of this element in $H_1$ has order $\leq \alpha! \beta! p^{\alpha-1} q^{\beta-1}(n-\alpha p - \beta q)!$, which will usually be smaller than $k!$, and you obtain a contradiction.

$\endgroup$
0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.