This problem is a follow up on my other MO question.

On the basis of experimental data, I'm prompted to ask:

Question. Let $R(a,b)$ an $a\times b$ rectangular grid, $h_{\square}$ the hook-length of a cell $\square$ in the Young diagram of a partition. Then,
$$\sum_{\mu\subset R(a,b)}\sum_{\square\in\mu}h_{\square}=\binom{a+b-2}{a-1}\binom{a+b+1}3.$$ Is this true? If so, any proof?

  • 5
    $\begingroup$ This looks like a problem about lattice paths in a rectangle, not so much about Young diagrams. After all, partitions that fit inside a rectangle are in bijection with lattice paths between two opposite corners of a rectangle. A hook can be subdivided into its arm and leg (and its source). I suspect that the sum of arms and the sum of legs can be computed separately. $\endgroup$ May 7, 2017 at 5:26
  • 1
    $\begingroup$ Yes, just replace the rhs by $\binom{a+b-3}{a-1}\binom{a+b}{3}$ $\endgroup$ May 7, 2017 at 8:27

2 Answers 2


As Darij suggests, we may count separately the sum of legs and the sum of arms. Denote the rows sizes $t_1,t_1+t_2,\dots,t_1+\dots+t_a$, where $t_i,1\leqslant i\leqslant a+1$, are non-negative integers and $\sum_{i=1}^{a+1} t_i=b$. Then the sum of arms (let's think that the arm includes the half of the source of the hook and the leg another half) equals $\sum_{i=1}^a \frac12(t_1+\dots+t_i)^2$. By symmetry, we know that the sums of $t_i^2$, $t_it_j$ over all diagrams do not depend on $i$, or respectively on a pair $i<j$. From $\sum t_i=b$ we see that the sum of $t_i$ (with fixed $i$) equals $\binom{a+b}b\cdot \frac{b}{a+1}$. Next, since $t_1^2+t_1t_2+\dots+t_1t_{a+1}=t_1b$, we see that $X+aY=\binom{a+b}b\cdot \frac{b}{a+1}$, where $X=\sum t_1^2$, $Y=\sum t_1t_2$. For finding $X$ we may observe that it is a coefficient of $x^b$ in $(\sum_{i=0}^\infty i^2x^i)(\sum_{i=0}^\infty x^i)^a$. We have $\sum_{i=0}^\infty i^2x^i=(x+x^2)(1-x)^{-3}$, so $X=[x^b](x+x^2)(1-x)^{-a-3}=\binom{a+b+1}{a+2}+\binom{a+b}{a+2}$. Rest is straightforward.


This is essentially lemma 2.5 in the paper AVERAGE SIZE OF A SELF-CONJUGATE (s, t)-CORE PARTITION.

  • $\begingroup$ Uhmm... No, it not, Harry. $\endgroup$ May 15, 2017 at 13:58

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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