2
$\begingroup$

By a distinct partition, I mean a partition into distinct parts, i.e., $10 = 5+4+1$ is one, but $10=6+2+2$ is not. The number of distinct partitions of $k$ all whose parts are at most $n$ is given by the sequence https://oeis.org/A053632. The link already provides the generating function as well as a recursive formula.

Now, I am interested in higher dimensional distinct partition. By an $q$-dimensional distinct partition of $k$, I mean a sum

$$k = \sum_{i_1,\cdots,i_q} \lambda_{i_1,\cdots,i_q}$$

for positive integers $\lambda_{i_1,\cdots,i_j,\cdots,i_q}$ such that $\lambda_{i_1,\cdots,i_q} < \lambda_{i_1,\cdots,i_j+1,\cdots,i_q}$ for every $j$.

What is known about the number of distinct $q$-dimensional partitions of $k$ into parts of size at most $n$?

Edit: Because that was maybe not clear: I'm aware that 1-, 2- and 3-dimensional partitions are ordinary, plane and solid partitions, respectively, and that there is an abundance of literature about them. However, I am interested in partitions into distinct parts as defined above. For example, the plane partition example on Wikipedia is not a partition into distinct parts. This case seems to be less covered, and I'm asking for literature about that.

$\endgroup$
8
  • 2
    $\begingroup$ If usual integer partitions are considered 1-dimensional, the 2-dimensional case is called "plane partitions" and a ton is known about them since the pioneering work of MacMahon: en.wikipedia.org/wiki/Plane_partition. The 3-dimensional case is called "solid partitions" and they are much more mysterious/badly behaved: en.wikipedia.org/wiki/Solid_partition. Higher dimensional partitions have been studied to a degree but not a lot is known. It's unclear what asymptotic regime you are interested in but hopefully these keywords give you a pointer to the literature. $\endgroup$ Commented Aug 9, 2024 at 12:28
  • $\begingroup$ @SamHopkins Yes, I'm aware of that. I edited the question to make clear that I want to count partitions into distinct parts. $\endgroup$ Commented Aug 9, 2024 at 17:11
  • 1
    $\begingroup$ Sorry, I missed that aspect of the question (and it wasn't clear to me you were aware of plane partitions, etc. from the way your question was originally phrased). For usual partitions there is an easy correspondence between partitions and partitions into distinct parts where we add 0 to the smallest part of a partition, 1 to the next smallest part, etc. You can try to do something similar with plane partitions where you add increasingly big numbers along anti-diagonals of the array, but I don't think it works out exactly the same as the 1D case. $\endgroup$ Commented Aug 9, 2024 at 17:21
  • $\begingroup$ Presumably each $i_j$ ranges over an interval $\{1,2,\dots,b\}$. Is $b$ allowed to depend on $j$ ? $\endgroup$ Commented Aug 9, 2024 at 18:00
  • 1
    $\begingroup$ @SamHopkins That's what I expected, in view of your comment about plane partitions, but I didn't see anything about it in the question. $\endgroup$ Commented Aug 9, 2024 at 18:06

1 Answer 1

6
$\begingroup$

For distinct part plane partitions (called strict plane partitions $\mathcal{SP}$ in the references below), MacMahon's formula

$$ \sum_{\pi \in \mathcal{P}} q^{|\pi|} = \prod_{n \ge 1} \left( \frac{1}{1-q^n} \right)^n$$

becomes

$$ \sum_{\pi \in \mathcal{SP}} 2^{k(\pi)}q^{|\pi|} = \prod_{n \ge 1} \left( \frac{1+q^n}{1-q^n} \right)^n$$

where $k(\pi)$ is "the number of connected components" of the plane partition $\pi$ as defined by Vuletić.

M. Vuletić, The Shifted Schur Process and Asymptotics of Large Random Strict Plane Partitions; Int. Math. Res. Notices (2007), Vol. 2007, article ID rnm043, 53 pages.

M. Vuletić, A Generalization of MacMahon's Formula; Trans. Am. Math. Soc. 361(5) (2009) 2789-2804.

These have both been cited several times, so looking through what referenced them might help with your higher dimensional questions.


As per Richard Stanley's and Sam Hopkins's comments, there's some tension between the question-writer's use of the term "distinct" and their definition (which allows for repeated parts across the entire partition). For plane partitions, I believe that's the difference between the OEIS sequences A117433 and A114736, respectively.

$\endgroup$
6
  • 1
    $\begingroup$ A "strict plane partition" in the sense of Vuletić, means a plane partition that is strictly increasing on diagonals, not that all parts are distinct. $\endgroup$ Commented Aug 10, 2024 at 14:33
  • 1
    $\begingroup$ @RichardStanley Thanks for the correction. $\endgroup$ Commented Aug 10, 2024 at 17:00
  • 1
    $\begingroup$ @RichardStanley But that is actually the condition the question-asker wants, even though they use the misleading term “distinct”! $\endgroup$ Commented Aug 10, 2024 at 17:28
  • 1
    $\begingroup$ @SamHopkins: doesn't the asker want strict increase in the coordinate directions? In two dimensions this is not the same as strict increase along the diagonals. $\endgroup$ Commented Aug 11, 2024 at 1:58
  • 1
    $\begingroup$ @RichardStanley You are right, I did not realize that this is what "strict plane partition" meant, so Brian Hopkins's answer is not really on point for what the question-asker wants. (But on the other hand, the question-asker also doesn't want all parts necessarily distinct.) $\endgroup$ Commented Aug 11, 2024 at 2:06

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.