Trying to understand a result in a representation theoretical paper, I realized that it implies the following elementary identity for symmetric functions. My question is whether this identity is true, known, and whether it has a direct proof. It might also happen that this identity is simpler than I think.

Let $d$ be a natural number. For $0\leq k\leq d$ let $e_k(z_1\dots,z_d)$ denote the $k$th elementary symmetric polynomial in $d$ variables, namely $$e_k(z_1\dots,z_d)=\sum_{1\leq i_1<\dots <i_k\leq d}z_{i_1}\dots z_{i_k}.$$

Let $x$ be a variable. The relevant identity is \begin{eqnarray*} \sum_{k=0}^d (-1)^k(x-d)^{2k}\cdot e_{d-k}((x-d+1)^2,\dots,(x-1)^2,x^2)=\\ d!\cdot (2x-2d+1)\dots (2x-d-1) (2x-d).\end{eqnarray*}

Notice that the right hand side is a polynomial in $x$ of degree $d$, while the left hand side has a priori degree at most $2d$ only.


1 Answer 1


Viete formula gives $\sum_{k=0}^{d} t^k e_{d-k}\left(\alpha_0,\ldots,\alpha_{d-1}\right)=(t+\alpha_1)\ldots(t+\alpha_d).$

Therefore (substituing $t=-(x-d)^2$ and $\alpha_j=(x-j)^2$), LHS is equal to $\prod_{j=0}^{d-1} \left(-(x-d)^2+(x-j)^2\right)= \prod_{j=0}^{d-1} (2x-j-d)(d-j)=d!\prod_{k=d}^{2d-1} (2x-k)$

which is exactly the RHS.


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.