0
$\begingroup$

$\newcommand\cd{\mathrm{cd}}$Let $A_m$ and $A_n$ be two alternating groups and $15\le m+2 \le n$. Denote $\cd_m$ and $\cd_n$ as the largest irreducible character degree of $A_m$ and $A_n$, respectively. I want to show that

$$ \frac{m!/2}{\cd_m} < \frac{n!/2}{\cd_n}. $$

My thought is to use upper and lower bounds for the largest irreducible character degrees of alternating groups. I've found a paper giving the upper bound and now I need the lower bound.

Given an alternating group $A_n$, denote its number of conjugacy classes as $k(n)$; is there a tighter lower bound for its largest irreducible character degree than $\sqrt{\frac{n!/2}{k(n)}}$?

$\endgroup$
2
  • 1
    $\begingroup$ The alternating group and symmetric group are not so different that I would imagine a maximizer would be similar, and so the vast literature on asymptotics of Young tableaux counting could be useful to you (e.g. I think it is known that a maximizer will be of Vershik-Kerov/Logan-Shepp shape). $\endgroup$ Jul 23, 2022 at 21:47
  • 2
    $\begingroup$ "Highest degree of an irreducible representation of the alternating group $A_n$" is tabulated at oeis.org/A060955 – for the symmetric group, see oeis.org/A003040 $\endgroup$ Jul 24, 2022 at 0:10

1 Answer 1

4
$\begingroup$

My original answer was flawed. I am replacing it with the following.

The irreducible representations $M_\lambda$ of the symmetric group $S_n$ are indexed by partitions $\lambda$ of $n$. If $\lambda\neq \lambda'$ (the conjugate partition to $\lambda$), then the restriction $N_\lambda$ of $M_\lambda$ to $A_n$ remains irreducible. If $\lambda=\lambda'$ then $M_\lambda$ splits into two irreps of the same dimension.

Now $n!/\dim(M_\lambda)$ is the product $H_\lambda$ of the hook lengths of $\lambda$. Thus $n!/(\dim N_\lambda)$ is either the product of the hook lengths (when $\lambda\neq \lambda'$) or twice this product. We are interested in those $\lambda$ that minimize $n!/(\dim N_\lambda)$. Suppose that $\lambda$ minimizes $n!/(\dim N_\lambda)$ and $\lambda=\lambda'$. Thus this minimum value is $2H_\lambda$. Remove any corner square from $\lambda$, giving a partition $\mu$ of $n-1$. Then clearly $H_\mu<2H_\mu<2H_\lambda$. Now suppose that $\lambda\neq\lambda'$. If $\mu=\mu'$ then we need to show $2H_\mu<H_\lambda$, which is not always true. However, it is easy to see that we can in fact remove a corner square from $\lambda$ so that $\mu\neq \mu'$ ($n\geq 3$), so $(n-1)!/(\dim N_\mu)= H_\mu<H_\lambda=n!/(\dim N_\lambda)$. Thus it seems to me that the numbers $(n!/2)/\mathrm{cd}_n$ are strictly increasing for $n\geq 3$.

$\endgroup$
2
  • $\begingroup$ This is very helpful. I was wondering if you could point out a reference for your fact about moving a box of a self-conjugate partition and getting a non-self-conjugate partition with less than double the hook length product? $\endgroup$
    – Suoria
    Jul 25, 2022 at 17:50
  • $\begingroup$ See corrected answer. $\endgroup$ Jul 31, 2022 at 19:31

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.