---
title: 'q-Narayana Numbers: Combinatorial & Algebraic Insights'
url: https://www.emergentmind.com/topics/q-narayana-numbers
type: topic
---

# q-Narayana Numbers: Combinatorial & Algebraic Insights

The $q$-Narayana numbers are a central class of $q$-analogues refining the $q$-Catalan numbers, interpolating between classical Catalan and Narayana sequences and their generalizations to higher types and parameterizations. They occur in enumerative combinatorics (especially lattice path and noncrossing partition statistics), symmetric functions, representation theory, and algebraic geometry. The $q$-Narayana numbers admit multiple parameterizations, possess rich algebraic properties (including $q$-$\gamma$-positivity, palindromy, and unimodality), satisfy convolution identities generalizing those of Kreweras and Le Jen-Shoo, and carry deep connections with cluster algebras, Hilbert series, and various combinatorial models.

## 1. Definitions and Explicit Formulas

The $q$-Narayana numbers have various definitions, depending on the context and type. The standard type-A $q$-Narayana numbers, denoted $N_q(n,k)$, are given by
\[
N_q(n,k) = \frac{1}{[n]_q} \binom{n}{k}_q \binom{n}{k-1}_q
\]
where $[n]_q = 1 + q + \cdots + q^{n-1}$ and $\binom{n}{k}_q = \frac{[n]_q!}{[k]_q! [n-k]_q!}$ is the $q$-binomial coefficient [1703.00003], [1605.09172].

A more general $q$-analogue, the $q$–$m$–Narayana numbers, is defined for $m \leq n$, $0 \leq k \leq n-m$:
\[
N^{(m)}_{n,k}(q)
= \frac{ {2m\brack m}_q }{ {n\brack m}_q }\, {n\brack k}_q\,{n\brack k+m}_q
\]
also expressible in $q$-Pochhammer notation [2505.09821]:

\[
N^{(m)}_{n,k}(q)
= \frac{ (q^{m+1};q)_m\,(q^{n-k+1};q)_k\,(q^{n-k-m+1};q)_{n+k} }
{ (q^{n-m+1};q)_m\,(q;q)_k\,(q;q)_{n+k} }
\]

These specialize at $m=0$ to type-B numbers of the form $N^{(0)}_{n,k}(q)$, e.g., $N_B(n, k; q) = \binom{n}{k}_q^2$ for the minimal nilpotent orbit case [2310.14302], [1605.09172].

Alternative formulations, such as the Hoggatt-type formula
\[
N_{n,k}(q) = \frac{1}{[k+1]_q} \binom{n}{k}_q \binom{n+1}{k}_q
\]
are commonly used in the Dyck path model [2601.08366].

For types $A$, $B$, and $C$ relative to Weyl groups (and at appropriate generalized parameters), Reiner and Sommers [1605.09172] provide:
- Type A$_{n-1}$, $\gcd(m,n)=1$: \
$\mathrm{Nar}(A_{n-1},m,k;q) = q^{(n-1-k)(m-1-k)} [k+1]_q \binom{n-1}{k}_q \binom{m-1}{k}_q$
- Types B$_n$, C$_n$, $m=2s+1$:
  \
$\mathrm{Nar}(B_{n},m,k;q)=\mathrm{Nar}(C_{n},m,k;q) = q^{2(n-k)(s-k)} \binom{n}{k}_{q^2} \binom{s}{n-k}_{q^2}$

## 2. Combinatorial and Representation-Theoretic Interpretations

At $q=1$, the (classical) Narayana numbers enumerate Dyck paths by number of peaks, noncrossing partitions by number of blocks, plane partitions in bounding boxes, and other combinatorial objects.

For general $q$, $N_q(n, k)$ refines the $q$-Catalan number by recording additional statistics (typically, area, major index, or other weight) on Dyck paths. The $q$–$m$–Narayana numbers refine the $q$–super Catalan numbers by the number of nonzero blocks in type-B noncrossing partitions, with $m=0$ yielding type-B Narayana numbers corresponding to noncrossing partitions of signed sets [2505.09821].

A specific combinatorial model arises from parallelogram polyominoes, where
\[
N_{m,n}(q,t) = \sum_{P \in PP(m, n)} q^{\area(P)} t^{\bounce(P)} = \sum_{P \in PP(m, n)} q^{\area(P)} t^{\dinv(P)}
\]
with statistics "area", "bounce", and $\dinv$ giving rise to the "bi-statistics" $q,t$-analogue and connections to diagonal harmonics and symmetric functions [1301.4803].

From a representation-theoretic perspective, the numerators of the $q$-Hilbert series of highest-weight coordinate rings (e.g., Grassmannians or minimal nilpotent orbits) are precisely $q$-Narayana polynomials [2310.14302]. For example, in type-A,
\[
H_{Gr(d, n+d+1)}(q, t) = \frac{\sum_{i=0}^{(d-1)n} N^{(q)}_A(d, n, i) t^i}{\prod_{r=1}^{d(n+1)} (1 - q^{r} t)}
\]
with each $N^{(q)}_A(d, n, i)$ encoding the graded multiplicity of $T$-weight spaces in Plücker coordinate rings.

For $q=-1$, $N_{n,k}(-1)$ counts symmetric Dyck paths by valleys [2601.08366].

## 3. Generating Functions, Recurrences, and $\gamma$-Positivity

For fixed $m, n$, the $q$-$m$-Narayana polynomials are
\[
\mathcal N^{(m)}_n(q; t) = \sum_{k=0}^{n-m} N^{(m)}_{n, k}(q)\, t^k
\]
These admit a $q$-$\gamma$-expansion:
\[
\mathcal N^{(m)}_n(q; t) = \sum_{s=0}^{\lfloor (n-m)/2 \rfloor} \left[{n-m \brack 2s}_q \, T_{m,s}(q)\right] t^s (1+t)^{n-m-2s}
\]
where $T_{m,s}(q)$ is the $q$-super Catalan number [2505.09821]. This expansion demonstrates $q$-$\gamma$-positivity, palindromy, and unimodality in $t$.

For type-A $q$-Narayana numbers, an explicit recurrence is:
\[
N^{(q)}_{n,k} = q^{n-k} N^{(q)}_{n-1,k} + N^{(q)}_{n-1, k-1}
\]
with boundary conditions $N^{(q)}_{n,0} = 1$, $N^{(q)}_{n,n} = 1$ [2310.14302].

For type-B:
\[
N_{B}(n, i; q) = \binom{n}{i}_q^2
\]
with the generating function $\prod_{j=0}^{n-1} (1 + u q^j)$ [2310.14302].

A bivariate generating function for the area- and bounce-refined $q$-Narayana numbers in the parallelogram polyomino model exhibits symmetry in $q$ and $t$ [1301.4803].

## 4. Convolution Identities and Symmetric Functions

Key convolution identities include the $q$-Kreweras and $q$-Le Jen-Shoo (Riordan-type) identities [2505.09821]:

- $q$-Kreweras:
\[
N^{(m)}_{n + k + m, k}(q) = \sum_{s=0}^k q^{(k-s)(k+m-s)} N^{(m)}_{n, k-s}(q) {2n + s \brack s}_q
\]
- $q$-Le Jen-Shoo:
\[
N^{(m)}_{n,k}(q) = \sum_{s=0}^{k-m} q^{s(s+m)} N^{(m)}_{k,s}(q) {n+k-s-m \brack 2k}_q
\]

These identities generalize classical convolution theorems to the $q$-Narayana setting, with proofs utilizing $q$-hypergeometric summations (${}_3\phi_2$) and $q$-Vandermonde techniques.

The $q$-Narayana polynomials are also expressible as moments of $q$-Fibonacci polynomials, linking them to continued fractions, inversion relations, and Hankel determinants [1611.05256].

In the symmetric-function framework, the $q, t$-Narayana polynomials correspond to inner products involving the nabla operator and complete symmetric functions:
\[
N_{m,n}(q, t) = (q t)^{m+n-1} \langle \nabla e_{m+n-2},\, h_{m-1} h_{n-1} \rangle
\]
giving a connection to bigraded Frobenius characteristics of diagonal harmonics [1301.4803].

## 5. Integrality, Positivity, and Specializations

The $q$-Narayana numbers are polynomials in $\mathbb N[q]$; negative powers of $q$ cancel via known $q$-Vandermonde and expansion formulas [2505.09821]. In all established expansions, coefficients are nonnegative integers.

Specializations produce various classical and combinatorial quantities:
- At $q=1$, ordinary Narayana numbers and Catalan numbers are recovered.
- At $q=-1$, $N_{n,k}(-1)$ counts symmetric Dyck paths with $k$ valleys [2601.08366].
- The $t=1$ specialization retrieves $q$-Catalan numbers:
\[
C_n(q) = \frac{1}{[n+1]_q} \binom{2n}{n}_q = \sum_{k=0}^n N_q(n, k)
\]
- At $t=1$, $N_{m,n}(q,1)$ provides $q$-refinements for ballot and Catalan numbers in polyomino and lattice-path models [1301.4803].

## 6. Advanced Examples and Open Problems

Recent research advances include:
- The relation of $q$-Narayana numerators to Hilbert series numerators for Grassmannians and minimal nilpotent adjoint orbits; in type-B, $N_B(n,i; q)$ encodes the $q$-enumeration of Grassmannian Schubert cells and appears in cluster-algebraic contexts [2310.14302].
- The cyclic sieving phenomenon for $q$-Narayana and $q$-Kreweras polynomials under the action of cyclic groups, with evaluations at roots of unity counting fixed points in rotational symmetry classes [1605.09172].
- Alternating sum congruences: for all positive integers $n$ and $r$, sums of powers of $q$-Narayana numbers modulo $q$-Catalan are always divisible by $C_n(q)$, a result with no known direct combinatorial proof [1703.00003].

Open problems remain, such as:
- Producing closed-form product formulas for the $q, t$-Narayana polynomials in the parallelogram polyomino model [1301.4803].
- Providing a bijective or representation-theoretic explanation for cluster/catalan coincidences in high types and orbits [2310.14302].
- Establishing combinatorial interpretations for the full range of $q$-Narayana statistics, especially for general convolution identities and specializations at roots of unity.

## 7. Comparative Table: Main Definitions

| Family                    | Formula                                                                                    | Reference         |
|---------------------------|--------------------------------------------------------------------------------------------|-------------------|
| Standard (type-A)         | $N_q(n, k) = \frac{1}{[n]_q} \binom{n}{k}_q \binom{n}{k-1}_q$                             | [1703.00003]      |
| Hoggatt (Dyck, peaks)     | $N_{n,k}(q) = \frac{1}{[k+1]_q}\binom{n}{k}_q\binom{n+1}{k}_q$                            | [2601.08366]      |
| $q$–$m$–Narayana          | $N^{(m)}_{n,k}(q) = \frac{ {2m\brack m}_q }{ {n\brack m}_q } {n\brack k}_q {n\brack k+m}_q$ | [2505.09821]      |
| Type-B                    | $N_B(n, k; q) = \binom{n}{k}_q^2$                                                         | [2310.14302]      |
| $q,t$-Narayana (polyomino)| $N_{m,n}(q,t) = \sum_{P} q^{\operatorname{area}(P)} t^{\operatorname{bounce}(P)}$         | [1301.4803]       |

The structure, recurrence, and combinatorial context of the $q$-Narayana numbers make them a deep, unifying object in algebraic combinatorics and its interactions with representation theory and algebraic geometry. Their many refinements, generalizations, and associated open problems continue to attract significant research attention.

Source: https://www.emergentmind.com/topics/q-narayana-numbers