# pattern-avoiding permutations vs multi-core partitions

Let $\mathfrak{S}_n$ be the permutation group on $[n]$. Given the pattern $\sigma=k(k-1)\cdots321$, let $I_n(\sigma)$ be the number of involutions in $\mathfrak{S}_n$ that avoid the pattern $\sigma$. Amitai Regev proved (see this paper and references therein) the case $k=4$: $$I_n((4321))=\sum_{k\geq0}\binom{n}{2k}C_k$$ which are the Motzkin numbers. Here $C_k$ are the Catalan numbers.

Given an integer partition $\lambda$, draw the Young diagram and fill in hook-lengths of each cell. If a number $t$ is not among these hook-lengths then $\lambda$ is called a $t$-core. If it misses $a, b, c$ then call it an $(a,b,c)$-core partition.

Let $N(n,n+1,n+2)$ be the number of all partitions that are $(n,n+1,n+2)$-core partitions. Then (see this paper for definitions and this result) $$N(n,n+1,n+2)=\sum_{k\geq0}\binom{n}{2k}C_k.$$

QUESTION. Is there a direct bijection between the above $(4321)$-avoiding involutions in $\mathfrak{S}_n$ and $(n,n+1,n+2)$-core partitions?

EDIT. Fedor's comment/question below alerted me to correct a mistake: we are only counting involutions and not all permutations.

• I find it reasonable to reproduce the definiton of $S$-core here. – Fedor Petrov Dec 12 '16 at 15:36
• I edited to account for this, hope it helps. – T. Amdeberhan Dec 12 '16 at 15:59
• Hm, what are six diagrams without hook lengths 3,4,5? – Fedor Petrov Dec 14 '16 at 17:50
• Well, empty diagram and three diagrams with 1 or 2 boxes, what else? – Fedor Petrov Dec 14 '16 at 17:53
• @Fedor: Well, the Motzkin numbers for $n\geq0$ are: $1, 1, 2, 4, 9, 21, 51, 127,\dots$. So, corresponding to $(3,4,5)$, there are $4$ core partitions. Where did you get $6$? – T. Amdeberhan Dec 14 '16 at 18:15