Fix $k,n\geq 1$. I wish to know the number of ways of dividing $n$ up as follows. First find $c_1,c_2,\ldots ,c_k$ such that $c_i\geq 0$ and $\sum_ic_i=n$. Then take each $c_i$ and find $c_{i,1},\ldots ,c_{i,n}$ such that $c_{i,m}\geq 0$ and $\sum_mmc_{i,m}=c_i$. How many ways are there to do this?
Asked
Viewed
261 times
1
$\begingroup$
$\endgroup$
-
1$\begingroup$ The "partitions" tag is a bit misleading, since, if I get you right, you don't want $c_1\geq c_2\geq ...$? $\endgroup$ – darij grinberg Aug 20 '10 at 14:02
-
2$\begingroup$ Anyway, generating functions should do the job. Denoting your number by $p_{n,k}$, we have $\left(\sum\limits_{\ell\in\mathbb N}\text{number of partitions of }\ell\cdot T^{\ell}\right)^k=\sum\limits_{n\in\mathbb N}p_{n,k}T^n$. $\endgroup$ – darij grinberg Aug 20 '10 at 14:05
-
2$\begingroup$ ... and this means $\left(\prod\limits_{u=1}^{\infty}\left(1-T^u\right)\right)^{-k}=\sum\limits_{n\in\mathbb N}p_{n,k}T^n$. $\endgroup$ – darij grinberg Aug 20 '10 at 14:06
-
$\begingroup$ Here is $k=2$: research.att.com/~njas/sequences/A000712 $\endgroup$ – darij grinberg Aug 20 '10 at 14:08
-
1$\begingroup$ The "representation theory" tag doesn't seem to be appropriate $\endgroup$ – Yemon Choi Aug 20 '10 at 23:55
