Take the 2-minute tour ×
MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

This is a cross-post from math.se, because I did not get any answer there:

Write $[n]:=\{1,\ldots,n\}$. For a partition $\lambda\vdash n$, I will write $[\lambda]$ for the Specht module that corresponds to $\lambda$, i.e. the complex vector space spanned by all standard tableaux of shape $\lambda$.

Consider the complex vector space $V_k$ which has as a basis the $k$-element subsets of $[n]$. The symmetric group $S_n$ acts on $V_k$ in the was that $g\in S_n$ maps a basis vector $A\subseteq[n]$ to the basis vector $g(A)=\{ g(i) \mid i\in A \}$. Let us assume $k\le \frac n2$. It is known that $$V_k = \bigoplus_{j=0}^k~ [n-j,j]$$ is the decomposition of $V_k$ into irreducibles. See, for instance, Example 7.18.8 in Stanley's book Enumerative Combinatorics (Volume 2).

Question: Given any $j\le\frac n2$ and a standard tableau $T$ of shape $\lambda=(n-j,j)$, I would like to know how to express the corresponding basis vector $v_T$ of $[\lambda]$ in the canonical basis of $k$-element subsets of $[n]$.

Thoughts: For $k=1$, this is rather easy. For the only standard tableau $T$ on $\lambda=(n)$, the vector $v_T=\sum_{i=1}^n \{ i \}$ is just the sum of all $1$-element subsets of $[n]$. On the other hand, for a standard tableau $T$ of $(n-1,1)$, let $a_T$ be the number in the single box in row $2$. Then it can be checked that $v_T=\{1\}-\{a_T\}$. However, I already have trouble with $V_2$. There is an obvious $S_n$-invariant inclusion \begin{align*} \phi_1: V_1 &\longrightarrow V_2 \\\\ \{i\} &\longmapsto \sum_{j\ne i} \{ i,j \} \end{align*} and the $[n-2,2]\subseteq V_2$ must be isomorphic to $V_2/\mathrm{im}(\phi_1)$, but I didn't get much further.

share|improve this question

1 Answer 1

Maybe I'm mistaken but I think the answer is quite easy if you use the language of Specht module, that is realize all these in polynomials. More precisely, you are working in the subspace of $C[x_1,\dots,x_n]$, where no variables are squared. In this space any standard tableau is realized as the product of the Vandermonde determinant of its columns. Then if I understand correctly your question, it just amount to expand such a product on the monomials. Does it makes sense ?

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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