10
$\begingroup$

Let $U$ be a representation of $S_m$ and $V$ a representation of $S_n$. Then the representation $\operatorname{Ind}_{S_m\wr S_n}^{S_{mn}}(U^{\otimes{n}}\otimes V)$ has a nice interpretation in terms of symmetric functions. If $\operatorname{ch}(U)$ and $\operatorname{ch}(V)$ are the Frobenius characteristics of $U$ and $V$ (symmetric functions of degree $m$ and $n$), then $$\operatorname{ch}\operatorname{Ind}_{S_m\wr S_n}^{S_{mn}}(U^{\otimes{n}}\otimes V)= \operatorname{ch}(V)[\operatorname{ch}(U)],$$ where the square brackets denote plethysm. I am interested in a graded version of this result.

More specifically, let $X$ be a space with an action of $S_m$ and $Y$ a space with an action of $S_n$. I am interested in the representation $$\operatorname{Ind}_{S_m\wr S_n}^{S_{mn}}\left(H^*(X^n)\otimes H^*(Y)\right)$$ of $S_{mn}$. The naive statement would be that $$\operatorname{ch}\operatorname{Ind}_{S_m\wr S_n}^{S_{mn}}\left(H^*(X^n)\otimes H^*(Y)\right) = \operatorname{ch}(H^*(Y))[\operatorname{ch}(H^*(X))],$$ where $\operatorname{ch}(H^*(X))$ and $\operatorname{ch}(H^*(Y))$ are symmetric functions of degree $m$ and $n$ with coefficients in the polynomial ring $\mathbb{Z}[t]$.

I'm pretty sure that this statement is correct when the cohomology of $X$ is concentrated in even degree, but otherwise it is wrong. The issue is that, at the level of $S_n$-representations, the Kunneth isomorphism $H^*(X^n)\cong H^*(X)^{\otimes n}$ needs to be interpreted in terms of super-vector spaces: the action of the simple transposition $(i,i+1)$ picks up a sign when the two cohomology classes being swapped have odd degree. For example, if $X=S^1$ and $n=2$, then $H^2(S^1\times S^1)\cong H^1(S^1)\otimes H^1(S^1)$ is the sign representation of $S_2$ rather than the trivial representation.

So my question is: Is there a "super version" of plethysm for symmetric functions with coefficients in $\mathbb{Z}[t]$ for which the last displayed equation is correct?

I know that the answer is tautologically "yes"--one can just translate from symmetric functions to representations, do the induction, and translate that, and take this as the definition of super-plethysm. But I'm looking for something explicit enough to allow me to do calculations in SAGE.

$\endgroup$
3
  • 1
    $\begingroup$ You just need to use an auxiliary variable $t$ for homological degree, and use the weight $(-t)^i$ for the $H^i$. The plethysm must act on this variable by $p_n(t)=t**n$. $\endgroup$
    – F. C.
    Mar 16, 2016 at 15:45
  • $\begingroup$ Ah, that makes sense, and it indeed seems to work in a simple example. Thanks! (If you want to post that as an answer, I'll accept it.) $\endgroup$ Mar 16, 2016 at 16:26
  • $\begingroup$ Does this provide an answer to my MO question mathoverflow.net/questions/33681/… with t=q? $\endgroup$ Mar 16, 2016 at 19:27

1 Answer 1

7
$\begingroup$

This amounts to study composition of "linear species" in the category of complexes.

The correct way to handle these computations using plethysm is to introduce an auxiliary variable $t$ and to weight the cohomology $H^i$ with the weight $(-t)^i$. The plethysm must act on $t$ by $p_n(t)=t^n$.

I do not know a written reference.

$\endgroup$
1
  • 2
    $\begingroup$ I was recently thinking about the same question. I've got a little bit different formula, but it is equivalent to the one above modulo the change of variables $q=-t$. If one introduces an auxiliary variable $q$ to weight the cohomology $H^i$ with the weight $q^i$, then the plethysm acts on $q$ by $p_n(q)=(-1)^{n-1}q^n$. $\endgroup$
    – Victor
    May 19, 2017 at 17:25

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.