Catalan Polyominoes: Structure & Enumeration

• Catalan polyominoes are column-convex shapes defined via Catalan words and Dyck paths, encoding key combinatorial properties and enumeration by Catalan numbers.
• They exhibit deep bijections with Dyck paths, parallelogram polyominoes, and path pairs, allowing refined statistics such as black cell capacity to be analyzed.
• Advanced techniques including generating functions, kernel methods, and continued fractions yield closed forms and succession rules, connecting them to Riordan arrays and generalized models.

Catalan polyominoes are column-convex polyominoes corresponding bijectively to Catalan words and, equivalently, Dyck paths, providing a fundamental combinatorial model encoding Catalan objects. Their structural characterization via sequences of column heights leads to deep enumerative connections, refined statistics, and rich bijective theory. Recent work enumerates statistics including black cell capacity using generating functions, functional equations, and continued fractions, and connects Catalan polyominoes to broader families such as parallelogram polyominoes, path pairs, and Riordan arrays.

1. Combinatorial Definition and Catalan Encoding

A Catalan polyomino of width $n$ is a column-convex polyomino constructed from a Catalan word $w = w_1w_2\cdots w_n$ of nonnegative integers satisfying $w_1=0$ and $w_i\leq w_{i-1}+1$ for $2\leq i\leq n$ (Baril et al., 24 Jan 2026). Each column $i$ has height $w_i + 1$, aligned on a common baseline. The set of all such polyominoes of width $n$, denoted $\mathcal{P}_n$, is in bijection with the set of Catalan words of length $n$; thus $|\mathcal{P}_n| = C_n$, the $n$th Catalan number.

This construction specializes classical parallelogram polyominoes, which are all regions enclosed between two non-intersecting north-east lattice paths from $(0,0)$ to $(k,\ell)$ with $k+\ell = n+1$ (Beaton et al., 2015). The Catalan polyomino class is strictly column-convex (bargraphs) and is the combinatorial core for numerous Catalan structures.

2. Bijections: Dyck Paths, Parallelogram Polyominoes, and Path Pair Models

Catalan polyominoes admit natural bijections:

• Dyck Path Bijection: The sequence of column heights $w$ encodes steps above or below a Dyck path; Catalan polyominoes are in bijection with Dyck paths of semilength $n$ (Beaton et al., 2015).
• Parallelogram Polyominoes: Catalan slicings of parallelogram polyominoes, where horizontal blocks have width 1, are counted by Catalan numbers via a succession rule (ECO method) (Beaton et al., 2015).
• Path Pairs Model: Parallelogram polyominoes correspond to path pairs $(\gamma_1,\gamma_2)$ in which the upper path $\gamma_1$ strictly stays above the lower path $\gamma_2$; for distance 1, these sets have cardinality $C_n$ (Drube, 2020).

Generalizations to Fuss–Catalan triangles via $k$–path pairs realize more exotic enumeration but the classical Catalan case ($k=2$) reduces to the standard bijections.

3. Refined Statistics: Black Cell Capacity and Vertical Measures

A chessboard coloring on Catalan polyominoes defines refined statistics, most notably black cell capacity $bc(P)$: cells are colored in a checkerboard pattern such that the southwestern cell is black, and $bc(P)$ is the total number of black cells (Baril et al., 24 Jan 2026). This statistic refines the area and is equidistributed (via a diagonal bijection) with the count of cells in odd-indexed columns (vertical black-cell capacity) and, depending on the parity of $n$, with vertical white-cell capacity. Theorem 1 of (Baril et al., 24 Jan 2026) establishes an involutive bijection $f$ on $\mathcal{P}_n$ such that, for odd $n$, $bc(P) = \operatorname{verblack}(f(P))$, and for even $n$, $bc(P) = \operatorname{verwhite}(f(P))$.

4. Enumerative Techniques: Functional Equations and Continued Fractions

The enumeration of Catalan polyominoes refines via multivariate generating functions incorporating width, last column height, and statistics such as black cell capacity: $$F(x,u,q) = \sum_{P\in\mathcal{P}} x^{\operatorname{length}(P)} u^{\operatorname{last}(P)} q^{bc(P)}$$ Splitting by parities yields a $4\times4$ system: $$\mathbf{F}(x,u,q) = \mathbf{M}(x,u,q)\,\mathbf{F}(x,\,q^{1/2}u,q) - \mathbf{N}(x,u,q)\,\mathbf{F}(x,1,q) + \mathbf{B}(x,u,q)$$ which can be solved by power series iteration (Baril et al., 24 Jan 2026).

Passing to Dyck paths and using automata, the generating functions satisfy linear recurrences of order 2; the solution gives a $2\times2$ matrix continued fraction whose inversion yields a closed form for the bivariate series $$G(x, q) = \sum_P x^{\operatorname{width}(P)} q^{bc(P)}$$ (Baril et al., 24 Jan 2026). Expanding: $$G(x, q) = xq + (q^2 + q)x^2 + (q^4 + 2q^3 + 2q^2)x^3 + \cdots$$ Further refinements generate vertical-capacity bivariate series $OD(x,y,z,u)$ and $EV(x, y, z, u)$, whose functional equations yield closed forms in terms of $q$-Pochhammer symbols.

5. Succession Rules, Generating Trees, and Kernel Method

Catalan slicings of parallelogram polyominoes grow under the succession rule "Cat": $$\textrm{root labeled }(1),\,\ (k)\ \longmapsto\ (1),\ (2),\ \dots,\ (k),\ (k+1)$$ Each Catalan slicing is labeled by the current rightmost column's height; vertical and horizontal block addition follows the rule. The generating tree produces the Catalan numbers at each depth (Beaton et al., 2015).

Formally, the corresponding bivariate series $A(x, u)$ satisfies: $$A(x,u) = x\,u + x\,u\,\frac{A(x,1)-A(x,u)}{1-u} + x\,u\,A(x,u)$$ The kernel method applies by solving $K(u) = 1-u+x\,u^2=0$, yielding $u = (1-\sqrt{1-4x})/(2x)$, which recovers the closed-form generating function $C(x)$ for Catalan objects.

6. Generalizations: Path Pair Families and Riordan Arrays

Catalan polyominoes situate in a family generalized by path pairs and Riordan arrays. For path pairs $(\gamma_1,\gamma_2)$ of equal lengths and prescribed intersection conditions, the number is counted by entries of Shapiro’s Catalan triangle. Extending to $k$–path pairs with divisibility conditions on vertical runs yields Riordan arrays $R(C_k(t)^{k-\varepsilon}, tC_k(t)^k)$, where $C_k(t)$ solves $C_k = 1 + t C_k^k$, and the enumeration follows closed forms in terms of Fuss–Catalan (Raney) numbers (Drube, 2020).

Weak path pairs admitting touching or crossing admit refinements by the number of returns, and all such models admit explicit closed forms via generating function and Lagrange inversion techniques.

7. Applications, Extracts, and Asymptotics

The explicit enumeration of Catalan polyominoes and statistics such as black cell capacity yields the univariate series $G(1, q) = \sum_P q^{bc(P)}$ with initial terms $2q + 5q^2 + 15q^3 + 47q^4 + 149q^5 + \cdots$ (Baril et al., 24 Jan 2026). For fixed width $n$, extraction of coefficients $[x^n] G(x, q)$ yields the distribution over $\mathcal{P}_n$. At specialization $q=1$, one recovers the ordinary generating series for Catalan polyominoes $\sum_n C_n x^n$; at $x=1$ the total black-capacity distribution is obtained. The functional equations admit further asymptotic analysis via kernel methods and saddle-point techniques, though explicit asymptotics for $[q^k]G(1, q)$ have not been derived in the cited work.

8. Proof Strategies and Analytical Tools

Enumeration and structural results on Catalan polyominoes employ a mix of:

• Column deletion and last-column analysis: Utilizing $w_i \leq w_{i-1}+1$ inequalities to decompose by last-column height.
• Parity refinement: Leading to low-dimensional matrix equations solvable by block-matrix inversion.
• Automata and continued fractions: Translating Dyck-path encodings to automaton recurrences, whose solution is represented as matrix continued fractions.
• Generating trees and kernel method: Applying recursive succession rules and algebraic kernel elimination for closed-form generating functions.
• Generalized Lagrange inversion: Extracting coefficients via explicit combinatorial formulas.

These methodologies unify advances on polyomino enumerations, lattice path models, and combinatorial algebraic structures (Baril et al., 24 Jan 2026, Beaton et al., 2015, Drube, 2020).