Tree-Rooted Planar Maps

Updated 28 December 2025

Tree-rooted planar maps are connected planar maps equipped with a distinguished spanning tree that encodes full combinatorial and topological information via blossoming trees and α‐orientations.
The methodologies incorporate bijections such as the Albenque–Poulalhon blossoming tree encoding and alternative schemes like Mullin’s code, providing linear-time algorithms for map reconstruction and sampling.
They contribute to generating function analysis, map enumeration, and scaling limit studies, offering insights into universality phenomena and efficient structural decompositions in random geometry.

A tree-rooted planar map is a connected planar map—i.e., a connected graph embedded in the sphere up to orientation-preserving homeomorphism—together with a distinguished spanning tree. This concept lies at the intersection of map enumeration, algebraic combinatorics, and bijective map theory, with deep connections to generating functions, orientational structures, and bijections to various classes of decorated trees. Such combinatorial structures serve as a unifying framework for encoding, enumerating, and sampling maps, and play a key role not only in pure map enumeration but also in scaling limits, random geometry, and universality phenomena.

1. Formal Definitions and Characterizations

Let $\mathcal{M}$ denote a combinatorial planar map and $T\subset \mathcal{M}$ a spanning tree. A tree-rooted planar map is the pair $(\mathcal{M}, T)$ with $\mathcal{M}$ rooted at a distinguished corner. Trees considered are always spanning (touching all vertices, acyclic, connected). In combinatorial conventions, trees are usually rooted and embedded, so the triplet $(\mathcal{M}, T, c_0)$ (where $c_0$ is the root corner) determines the isomorphism class.

A central perspective is in terms of blossoming trees. Here, the map–tree pair is encoded as a plane tree $T$ decorated with stems—dangling half-edges classified into “openings” and “closings”—that encode exactly the non-tree edges in $\mathcal{M}$ . The cyclic order of edges and stems inherited from the embedding captures the full combinatorial and topological information needed to reconstruct the map (Albenque et al., 2013).

In higher genus, the same setup generalizes to so-called quasi-trees, but in the planar (genus 0) case, quasi-trees are precisely spanning trees (Cori et al., 2022).

2. Bijections: Blossoming Trees and Beyond

Blossoming Tree Encoding

The generic bijection introduced by Albenque and Poulalhon encodes any tree-rooted planar map $(M, T)$ as a blossoming tree, by deleting all non-tree edges and replacing each with an “opening” and a “closing” stem attached at each endpoint. The outer-face contour walk prescribes how to reconstruct $\mathcal{M}$ : each opening, immediately followed by a closing, can be uniquely paired to recover the deleted edge. The closure operation iterates these pairings until the original map is restored (Albenque et al., 2013).

Canonical Orientations and $\alpha$ -Orientations

To achieve canonical and algorithmically efficient bijections—independent of the chosen spanning tree—this theory invokes $\alpha$ -orientations (Felsner). For a feasible function $\alpha:V(\mathcal{M})\rightarrow\mathbb{N}$ , a minimal accessible $\alpha$ -orientation determines a unique tree–cycle partition: tree-edges and closure-edges. The bijection specializes to various families of maps (Eulerian, nonseparable, $d$ -angulations, etc.) by the appropriate choice of $\alpha$ (Albenque et al., 2013).

Generalizations and Alternative Encodings

Alternative perspectives—Mullin’s code, Lehman-Lenormand code, fighting fish, and double occurrence words—provide further bijections with walks in $\mathbb{N}^2$ , generalized polyominoes, or bicolored chord diagrams. For instance, the Mullin code encodes $(\mathcal{M},T)$ as a 2n-step walk in $\mathbb{N}^2$ , and is in bijection with the class of generalized fighting fish (Duchi et al., 2022).

Additionally, explicit combinatorial correspondences exist between tree-rooted maps and bipartite circle graphs, chord diagrams, and non-crossing matchings (Cori et al., 2022, Bousquet-Mélou et al., 21 Dec 2025).

3. Enumeration Formulas and Generating Functions

The enumerative theory of tree-rooted planar maps has a rich structure, with explicit closed forms and generating functions for various subclasses.

Mullin’s formula: The number of tree-rooted maps with $n$ edges is

$t_n = \frac{2\cdot3^n}{(n+1)(n+2)}\binom{2n}{n}$

which also counts rooted planar maps with $n$ edges (Albenque et al., 2013).

For prescribed vertices and faces:

$|\mathrm{TM}(i,j)| = \frac1{(i+1)(j+1)} \binom{2i+2j}{i,i,j,j}$

(Bousquet-Mélou et al., 21 Dec 2025).

For tree-rooted planar cubic maps (each vertex degree three, with a marked spanning tree and a directed non-tree edge), the count is

$C_{2n}\cdot C_{n+1}$

with $C_k=\frac{1}{k+1}\binom{2k}{k}$ the Catalan number, and the generating function $M(z)$ satisfies a quadratic equation. Removing the root-edge marking yields more involved, piecewise explicit formulas via Burnside’s lemma (Kochetkov, 2016, Kochetkov, 2017).

Generating functions for tree-rooted planar maps also satisfy quadratic or higher-order algebraic equations, amenable to analysis by Lagrange inversion and singularity analysis (Cori et al., 2022).

4. Connections to Map Classes and Structural Decompositions

The generic bijective framework encompasses all major classical families:

Eulerian maps: The Albenque–Poulalhon bijection recovers Schaeffer’s construction via $\alpha(v) = \deg(v)/2$ ; the closure step yields mobiles with prescribed labels.
m-constellations: By coloring and degree assignments ( $\alpha$ function) one captures Bousquet-Schaeffer’s m-ary mobiles.
Nonseparable maps: Specialization to quadrangulations with $\alpha(v)=2$ or via the Mullin code (rightmost DFS spanning tree) recovers the Brown–Tutte–Schaeffer–Jacquard formula.
Simple triangulations/quadrangulations: Via $\alpha=1$ or $2$ at inner vertices, reproducing Poulalhon–Schaeffer’s bijections.
Block decompositions: Tree-rooted maps decompose into trees of 2-connected blocks, allowing for fine-grained enumerative and probabilistic analysis, and capturing the geometry of the “block tree” (Albenque et al., 2024).

The blossoming-tree framework unifies the encoding of maps from all these families via $\alpha$ -orientations, and specialized coding schemes exist for higher-connectivity or simple maps (Fusy, 28 Oct 2025).

5. Probabilistic and Scaling Properties

Tree-rooted planar maps exhibit rich phase transitions when weighted by structural features, such as the number of 2-connected blocks. Weighting by a factor $u$ per block, the system undergoes a phase transition at a critical value $u_C\approx3.02$ :

Subcritical ( $u < u_C$ ): Maps typically have a giant 2-connected block of linear size, with remaining blocks sublinear.
Critical ( $u = u_C$ ): Largest blocks are of size $\Theta(\sqrt{n})$ , and the map metric exhibits a $\sqrt{n/\log n}$ scaling.
Supercritical ( $u > u_C$ ): All blocks are logarithmic in size, and the scaling limit is the Brownian Continuum Random Tree (CRT) (Albenque et al., 2024).

Such results are established via functional equations on generating functions, singularity analysis, and probabilistic coupling to conditioned Galton–Watson trees. The scaling limit (in critical/supercritical regimes) is rigorously Brownian CRT, and thus tree-rooted map models are key to the conjectured universality of the Brownian map for random planar maps (Albenque et al., 2024).

6. Algorithms, Encodings, and Applications

Generic bijections induce efficient, linear-time encoding and sampling algorithms: the closure and opening procedures on blossoming trees, contour codings, or fighting-fish evolutions all have linear complexity in reasonable map subclasses. The resulting encodings allow random sampling with optimal entropy (Albenque et al., 2013, Fusy, 28 Oct 2025). For example, 5-connected triangulations can be encoded as leg-balanced trees with bijections to algebraic generating functions and succinct word representations (Fusy, 28 Oct 2025).

Tree-rooted planar maps, through their encodings, also serve as combinatorial backbones for cyclic sieving phenomena (CSP). Natural cyclic actions (e.g., rotating the root corner around the tree) yield CSP polynomials in terms of $q$ -binomials and multinomials, providing deep connections between map enumeration, representation theory, and symmetric functions (Bousquet-Mélou et al., 21 Dec 2025).

7. Structural Insights and Universal Applications

The versatility of the tree-rooted model extends to bijections with other combinatorial objects: circle graphs, chord diagrams, non-crossing matchings, double occurrence words, and generalized fighting fish (Cori et al., 2022, Duchi et al., 2022). Their encoding power is central both to analytic combinatorics (generating functions and asymptotics) and to explicit geometry (layer decompositions, shortest-path metrics, and scaling limits).

The framework establishes not only a unifying technical approach to map enumeration but also practical tools for the study of random topological surfaces, quantum gravity (via Brownian map universality), and phase transition phenomena in complex random structures (Albenque et al., 2024). Tree-rooted planar maps thus serve as a central combinatorial paradigm linking bijective, probabilistic, and geometric aspects of map theory and related fields.