MathOverflow will be down for maintenance for approximately 3 hours, starting Monday evening (06/24/2013) at approximately 9:00 PM Eastern time (UTC-4).

## Higher-dimensional Catalan numbers?

One could imagine defining various notions of higher-dimensional Catalan numbers, by generalizing objects they count. For example, because the Catalan numbers count the triangulations of convex polygons, one could count the tetrahedralizations of convex polyhedra, or more generally, triangulations of polytopes. Perhaps the number of triangulations of the $n$-cube are similar to the Catalan numbers? Or, because the Catalan numbers count the number of below-diagonal monotonic paths in an $n \times n$ grid, one could count the number of monotonic paths below the "diagonal hyperplane" in an $n \times \cdots \times n$ grid.

I would be interested in references to such generalizations, which surely have been considered—I am just not looking under the right terminology. I would be particularly grateful to learn of generalizations that share some of the Catalan number's ubiquity. Thanks for pointers!

-
What do you mean by diagonal hyperplane? – Gjergji Zaimi Nov 21 2011 at 1:56
@Gjergji: One generalization of $y \le x$ would be $x_d \le x_1 + x_2 + \cdots x_{d-1}$. But other generalizations might be more natural... – Joseph O'Rourke Nov 21 2011 at 2:10

When you count the number of positively directed paths from $(0,0,\dots,0)$ to $(n,n,\dots,n)$ that lie in the region $x_d\le x_1+\cdots+x_{d-1}$, you can project to the plane $(x_d,x_1+\cdots+x_{d-1})$ and find that you need the number of planar paths from $(0,0)$ to $(n,n(d-1))$ which stay above the line $x=y$, and which have $n$ vertical steps of each color {1,...,d-1}. So the answer comes to be $$\left(\binom{nd}{n}-\binom{nd}{n-1}\right)\binom{n(d-1)}{n,\dots,n}=\frac{n(d-2)+1}{n(d-1)+1}\frac{(nd)!}{(n!)^d}.$$ See also this previous question for enumerating lattice paths below a line.

On the other hand, one way to interpret Catalan numbers as lattice paths below the diagonal is to look at it as counting the number of standard Young tableaux of shape $(n,n)$. So a natural generalization is for example the number of standard Young tableaux of shape $(n,n,\dots,n)$. This corresponds to the region $x_1\le x_2\le\cdots\le x_d$, and can be counted using the hook-length formula. See my answer here.

-
 Also see oeis.org/A005789 – domotorp Nov 21 2011 at 9:25 @Gjergji: Impressive computation for the number of monotone paths! @domotorp: Thank. For the record, the OEIS sequences is the "Number of standard tableaux of shape $(n,n,n)$." – Joseph O'Rourke Nov 21 2011 at 10:56

(1) Free n-category generated by a cube, oriented matroids, and higher Bruhat orders by Karpranov and Voevodsky and

(2) Arrangements of hyperplanes, higher braid groups and higher Bruhat orders by Manin and Schechtman

Maps between higher Bruhat orders and higher Stasheff-Tamari posets by H. Thomas serves as a follow-up and fills in in some of the details, but has a different flavor.

Briefly, paper (1) defines (among other things) a poset $S(n,k), 0 \leq k \leq n$, with $|S(n,2)|$ equal to the $n^{th}$ Catalan number. The objects of this poset are defined in 3 ways (and shown to be equivalent): combinatorially (via cyclic polytopes), as "homotopies of homotopies" and as elements of "the free $(n-k)$-category on the $n$-simplex".

So if you like these definitions and think they are natural, $|S(n,k)|$ for $k >2$ gives you "higher dimensional" Catalan numbers.

Paper (2) does something nice, too, but I can't remember what.

-

This is not really an answer, but I'm just wondering if there are generalizations (in $2$-dimensional setting) of the form: number of paths from $(0,0)$ to $\left(n,\left\lfloor mn \right\rfloor\right)$ that stay below the line $y=mx$? And would that have other interpretations?

-
 This paper discusses a related problem (Generalized ballot problem): thales.math.uqam.ca/~serrano/paths.pdf – Vasu vineet Nov 22 2011 at 5:52

I am interested in nested sequences of Dyck paths. I rotate your picture and think of a Dyck path as a sequence of steps $(1,1)$ and $(1,-1)$ which stays above $y=0$. A nested sequence is a sequence of $k$ Dyck paths such that if $1\le i<j\le k$ then the $i$-th path is never above the $j$-th path. I don't know of any interpretation in terms of polytopes.

-

The closest thing I can think of to what you want is triangulations of even dimensional cyclic polytopes. These are nice because, unlike triangulations of the cube, they all use the same number of simplices. See the above linked paper for more.

There are a number of other important generalizations of Catalan numbers, but these are the only ones I would particularly call higher-dimensional.

-
 Thank you, David! I would never have thought to look in this direction: "Higher dimensional cluster combinatorics and representation theory," by Steffen Oppermann and Hugh Thomas. – Joseph O'Rourke Nov 21 2011 at 1:59