betaFeedback

WorksMathematics

Bijecting mm-Tamari intervals, trees, and tight gradings

Abstract

Tight gradings, which are certain integer gradings on maximal Dyck paths, provide a combinatorial framework for computing all coefficients of wall-functions in rank-22 cluster scattering diagrams. We construct a bijection between a special subset of tight gradings and the intervals of the mm-Tamari lattice, providing a new combinatorial proof of the central wall-function formula proved by Reineke. Inspired by the structure of these gradings, we introduce a class of plane trees called mm-leafy trees that give a new model for mm-Tamari intervals. The defining condition of an mm-leafy tree is simple: in every full subtree, the number of leaves is greater than mm times the number of internal vertices. The leafy tree perspective gives rise to a quadratic functional equation with a catalytic parameter, providing an alternative way to enumerate mm-Tamari intervals.

Introduction

The Tamari lattice Tn\mathcal{T}_n, introduced by Tamari in 1951, is the partial order on binary trees with nn internal vertices generated by rotations. Its Hasse diagram is an orientation of the edge graph of an associahedron, linking Catalan combinatorics with the geometry of associativity. Tamari intervals, namely comparable pairs in Tn\mathcal{T}_n, gained sustained interest after Chapoton showed in 2006 [10] that they are counted by 2n(n+1)(4n+1n−1)\frac{2}{n(n+1)}\binom{4n+1}{n-1}. This simple formula inspired the development of several combinatorial models, for example [3, 11, 12].

Bergeron observed that Chapoton’s formula coincides with Haiman’s conjectural dimension of the alternating component of the SnS_n-module of trivariate diagonal harmonics. Motivated by mm-variant versions of these spaces, he introduced the mm-Tamari lattice Tnm\mathcal{T}^{m}_n [2] and conjectured that its intervals are counted by

m+1n(mn+1)((m+1)2n+mn−1).(1)\frac{m+1}{n(mn+1)}\binom{(m+1)^2n+m}{n-1}. \tag*{(1)}

Bousquet-Mélou, Fusy, and Préville-Ratelle subsequently proved this formula counting the mm-Tamari intervals [5].

More recently Chapoton observed that the numbers (1) also coincide with the Euler characteristics of a family of Kronecker moduli spaces [1], which were computed by Weist [21]. These Kronecker moduli are known to be closely related to scattering diagrams in rank 2 [20, 15, 9].

Our approach to mm-Tamari intervals arises from the computation of cluster scattering diagrams [14]. In prior work [8], the authors introduced tight gradings, which provide a uniform combinatorial formula for wall-function coefficients in any rank-2 cluster scattering diagram. These gradings on Dyck paths are built on Lee–Li–Zelevinsky’s compatible pairs [16], and play a crucial role in extending Laurent positivity to generalized cluster algebras [7]. While seeking a combinatorial enumeration of tight gradings associated with the central wall, we noticed that the number of 0-framed tight gradings is precisely that of mm-Tamari intervals. This equinumerosity motivated us to look for an explicit combinatorial connection.

To establish this connection, we introduce mm-leafy trees as a novel plane tree model for mm-Tamari intervals. Their defining condition is surprisingly simple: in every full rooted subtree, the number of leaves is strictly larger than mm times the number of internal vertices. These trees serve as intermediate objects bridging mm-Tamari intervals and tight gradings.

The main aim of this extended abstract is to present the bijections between these three sets of combinatorial objects:

  • (I) intervals in the mm-Tamari lattice of order nn,

  • (II) mm-leafy trees on nn internal vertices and mn+1mn+1 leaves, and

  • (III) 0-framed tight gradings contributing to the nnth central wall coefficient of the cluster scattering diagram Dm+2,m+2\mathcal{D}_{m+2,m+2}.

The corresponding mm-Tamari interval in (I) can be read directly from the right-to-left preorder of a leafy tree in (II): the maximum element records whether each vertex is internal or a leaf, while the minimum element records the arities of the internal vertices. Moreover, we extend the bijection between (II) and (III) to a bijection between mm-leafy trees with a distinguished vertex and the set of all tight gradings contributing to the nnth central wall coefficient of Dm+2,m+2\mathcal{D}_{m+2,m+2}.

The bijection to mm-leafy trees yields several nice consequences for mm-Tamari intervals. The recursive structure of leafy trees makes them easy to generate and enumerate, and they yield a comparability test for mm-Tamari elements as a positivity condition on plane trees. We find that a generating series for mm-leafy trees, refined by two tree statistics, satisfies the functional equation (7) obtained in [5] for mm-Tamari intervals. We also derive a new quadratic functional equation (10) with one catalytic parameter using the tree structure. Solving this equation gives an alternative simpler derivation of the formula (1). The obvious left-right symmetry on mm-leafy trees induces an involution on mm-Tamari intervals which was not previously identified, to the best of our knowledge.

The connection with tight gradings gives a second application. Although tight gradings are proven to compute wall coefficients [7], we had not been able to combinatorially enumerate them, even for the central wall, until linking them with leafy trees. So the tree perspective provides a new proof of the central wall formula for Dm+2,m+2\mathcal{D}_{m+2,m+2}, conjectured by Kontsevich in a closely related setting of tropical vertex [15] and proved by Reineke using quiver representations [19, 9].

A tree model for mm-Tamari intervals

In this section, we review the mm-Tamari lattice and its intervals, introduce mm-leafy trees, and describe the bijection Φ\Phi from mm-leafy trees to mm-Tamari intervals.

Intervals in the mm-Tamari lattice

Dyck paths of size nn form the Tamari lattice Tn\mathcal{T}_n. An mm-ballot path of size nn is a word with nn letters NN and mnmn letters EE such that every prefix WW satisfies m∣W∣N−∣W∣E≥0m|W|_N-|W|_E \ge0, where ∣W∣L|W|_L is the number of letter LL in WW. Equivalently, it is a north-east lattice path from (0,0)(0,0) to (mn,n)(mn,n) that stays non-strictly above x=myx=my. Thus mm-ballot paths generalize Dyck paths where m=1m=1.

Definition 2.1. For PP and QQ two mm-ballot paths of size nn, define P≺QP \prec Q if there exists in PP an east step aa, followed by a north step bb, such that QQ is obtained from PP by swapping aa and SS, where SS is the shortest factor of PP that begins with bb and is a (translated) mm-ballot path.

The transitive closure ≤\le of ≺\prec defines a lattice on mm-ballot paths of size nn, which we call the mm-Tamari lattice Tnm\mathcal{T}_n^m, generalizing T11=Tn\mathcal{T}_1^1=\mathcal{T}_n. Any pair P,Q∈TnmP,Q\in\mathcal{T}_n^m with P≤QP\le Q is an interval and we denote it by [P,Q][P,Q]. Write Inm\mathcal{I}_n^m for the set of all intervals.

Figure 1. For the Hasse diagram of the Tamari lattice T3=T31\mathcal{T}_3=\mathcal{T}_3^1 (left), we depict the leafy trees (center) and tight gradings (right) corresponding to the vertices and edges. The intervals of size greater than 2 are described below.

We refer the reader to [5] for a more comprehensive exposition on mm-Tamari lattices and their intervals. We note that there have been combinatorial models for mm-Tamari intervals, for example [11, 18], and it would be interesting to understand the relation with our tree model introduced below.

m-leafy trees

All trees considered here are finite plane trees; in particular, there is a designated root and the children of each internal vertex (i.e. non-leaf vertex) are ordered from left to right. In our illustrations, the root is always placed at the top.

A subtree of a tree TT is a plane tree rooted at a vertex vv of TT spanned by all descendants of vv, and is denoted by TvT_v.

Let ℓ(T)\ell(T) denote the number of leaves of TT and n(T)n(T) the number of internal vertices.

Fix a positive integer mm. The following notion of mm-leafy trees is central to our discussion but remarkably simple.

Definition 2.2 (mm-leafy trees). A tree TT is called mm-leafy if for every vertex vv,

χm(Tv):=ℓ(Tv)−mn(Tv)>0.\chi_m(T_v) := \ell(T_v) - mn(T_v) > 0.

Despite this simple characterization, the class of mm-leafy trees seems not to have drawn much attention in combinatorics. However, some special cases are familiar families. For example, mm-leafy trees with χm(Tv)=1\chi_m(T_v)=1 for every vv are precisely (m+1)(m+1)-ary trees.

A 2-leafy tree with values of $\chi_2(T_v)$ next to each vertex

Figure 2. A 2-leafy tree with χ2(Tv)\chi_2(T_v) next to each vertex vv.

For n≥1n \ge1, write LTnm:={T:T is m-leafy with n internal vertices and χm(T)=1}\mathrm{LT}^{m}_{n}:=\{T:T\text{ is }m\text{-leafy with }n\text{ internal vertices and }\chi_m(T)=1\}. Equivalently, LTnm\mathrm{LT}^{m}_{n} is the set of mm-leafy trees with nn internal vertices and mn+1mn+1 leaves.

Bijecting m-leafy trees and m-Tamari intervals

The promised bijection Φ:LTnm→Inm\Phi:\mathrm{LT}^{m}_{n}\to\mathcal{I}^{m}_{n} is easy to describe.

Let SS be a tree. The right-to-left preorder of SS is defined to visit the root first, then recursively visit the children from right to left. List among all vertices the internal vertices in this order v1,v2,…,vnv_1,v_2,\ldots,v_n. Let kik_i be the arity of viv_i, the number of children of viv_i.

We associated the upper and lower words to SS:

US:=the word of all vertices in the right-to-left preorder,with N for internal vertices and E for leaves, after deleting the final E;(2)U_S := \text{the word of all vertices in the right-to-left preorder,}\\ \text{with }N\text{ for internal vertices and }E\text{ for leaves, after deleting the final }E; \tag*{(2)}
LS:=NEk1−1NEk2−1⋯NEkn−1.(3)L_S := NE^{k_1-1}NE^{k_2-1}\cdots NE^{k_n-1}. \tag*{(3)}
A 2-leafy tree in $\mathrm{LT}^{2}_{3}$ and the associated words

Figure 3. A 2-leafy tree in LT32\mathrm{LT}^{2}_{3} and the associated words.

Theorem 2.5. If S∈LTnmS \in\mathrm{LT}^{m}_{n}, then LSL_{S} and USU_{S} are mm-ballot paths with LS≤USL_{S} \leq U_{S}. Moreover, the map below is a bijection:

Φ:LTnm→Inm,Φ(S):=[LS,US].(4)\Phi:\mathrm{LT}^{m}_{n} \to\mathcal{I}^{m}_{n}, \qquad\Phi(S):=[L_{S},U_{S}]. \tag*{(4)}

Next we discuss some consequences of the above tree model of mm-Tamari intervals.

A plane tree is called a caterpillar if its internal vertices form a chain. A caterpillar is called a left (resp. right) caterpillar if any internal vertex is the leftmost (resp. rightmost) child of its parent.

The bijection Φ\Phi gives rise to three models for the mm-Tamari lattice:

  • Left mm-leafy caterpillars are length-zero intervals, namely mm-ballot paths;

  • Right mm-leafy caterpillars are intervals whose upper path is maximal;

  • (m+1)(m+1)-ary trees are intervals whose lower path is minimal.

A left caterpillar, a right caterpillar, and a ternary tree

Figure 4. A left caterpillar, a right caterpillar, and a ternary tree corresponding to the same 2-ballot path NENENEEEEENEENENENEEEEENEE.

The third model is a well-known (m+1)(m+1)-ary tree description of the mm-Tamari lattice. The cover relation is realized as certain tree rotation, which was already pointed out in [5 Section 4].

We may characterize the mm-leafy trees that correspond to cover relations in Tnm\mathcal{T}^{m}_{n}. See Figure 5 for an example.

A 2-leafy tree $S$ that is a cover relation

Figure 5. A 2-leafy tree SS that is a cover relation.

Proposition 2.7. For S∈LTnmS \in\mathrm{LT}^{m}_{n}, the interval Φ(S)=[LS,US]∈Inm\Phi(S)=[L_{S},U_{S}] \in\mathcal{I}^{m}_{n} is a cover relation if and only if the two conditions hold:

  • SS has precisely one internal vertex vv that is the second leftmost child of its parent, and any other non-root internal vertex is the leftmost child of its parent.

  • χm(Sv)<χm(Sv′)\chi_{m}(S_{v}) < \chi_{m}(S_{v'}) for any internal descendant v′v' of vv.

The left-right reflection of plane trees defines an involution τ:LTnm→LTnm\tau: \mathrm{LT}^{m}_{n} \to\mathrm{LT}^{m}_{n}. In particular, it sends a left caterpillar to its mirrored right caterpillar, and vice versa. The set of (m+1)(m + 1)-ary trees is closed under τ\tau.

The involution τ\tau gives rise to the involution on mm-Tamari intervals

γm:=Φ∘τ∘(Φ)−1:Inm→Inm,\gamma_m := \Phi\circ\tau\circ(\Phi)^{-1} : \mathcal{I}^{m}_{n} \to\mathcal{I}^{m}_{n},

which exchanges length-zero intervals and upper intervals. We remark that γm\gamma_m is different from the involution βm\beta_m constructed by Pons since βm\beta_m sends length-zero intervals to length-zero ones [18 (5.36)].

The last page after the bibliography illustrates Theorem 2.5 for Φ:LT32→I32\Phi: \mathrm{LT}^{2}_{3} \to\mathcal{I}^{2}_{3}, where 2-ballot paths are represented by left 2-leafy caterpillars.

Enumeration of mm-leafy trees

The bijection Φ:LTnm→Inm\Phi: \mathrm{LT}^{m}_{n} \to\mathcal{I}^{m}_{n} already implies the closed-form expression

∣LTnm∣=m+1n(mn+1)((m+1)2n+mn−1),(5)\left|\mathrm{LT}^{m}_{n}\right| = \frac{m+1}{n(mn+1)}\binom{(m+1)^2n+m}{n-1}, \tag*{(5)}

as ∣Inm∣\left|\mathcal{I}^{m}_{n}\right| equals the same expression (1).

Trees are among typical combinatorial objects that can be constructed recursively based on disjoint union and concatenation. This feature makes it possible to write down functional equations of their generating series. We present two functional equations involving generating series of mm-leafy trees. The first already appeared in the enumeration of mm-Tamari intervals [5]. The second appears to be new and can be solved following general methods developed in [6]; see also [4].

First functional equation

For a tree TT, let l(T)l(T) (resp. r(T)r(T)) be the number of internal vertices in the longest chain obtained by taking only the leftmost (resp. rightmost) child starting from the root. For a single leaf, l=r=0l = r = 0. Consider the generating function

F(m)(t;x,y):=∑n≥0∑T∈LTnmtnxl(T)+1yr(T).(6)F^{(m)}(t; x, y) := \sum_{n \ge0}\sum_{T \in\mathrm{LT}^{m}_{n}} t^n x^{l(T)+1} y^{r(T)}. \tag*{(6)}

Proposition 3.3. Writing F(x,y)≡F(m)(t;x,y)F(x,y) \equiv F^{(m)}(t;x,y), we have

F(x,y)=x+xyt(F(x,1)⋅Δ)mF(x,y),(7)F(x,y)=x+xyt\bigl(F(x,1)\cdot\Delta\bigr)^m F(x,y), \tag*{(7)}

where Δ\Delta is the divided difference operator ΔS(x)=S(x)−S(1)x−1\Delta S(x)=\frac{S(x)-S(1)}{x-1}.

The equation (7) is precisely [5 (3)], although interpreted in terms of mm-Tamari intervals using the initial rise of upper path and the contact of lower path. We remark that via the bijection Φ\Phi, the initial rise translates to our rr whereas the contact does not translate to our ll. In fact, the quantity l(T)l(T) is the number of north steps shared by the mm-ballot paths LTL_T and UTU_T, while r(T)r(T) is the number of initial north steps of UTU_T.

The tree interpretation makes the following symmetry (described as a combinatorial mystery for rise-contact in [5]) evident, since the mirror map τ\tau exchanges the statistics ll and rr:

F(m)(t;x,y)/x=F(m)(t;y,x)/y.(8)F^{(m)}(t;x,y)/x=F^{(m)}(t;y,x)/y. \tag*{(8)}

Second functional equation

While the solution to (7) in [5] is rather complex, we present another functional equation with a catalytic parameter uu whose solution is simpler. Define the generating series

G(m)(t,u)=1+∑Ttn(T)uχm(T)∈Q[[t,u]],(9)G^{(m)}(t,u)=1+\sum_T t^{n(T)}u^{\chi_m(T)}\in\mathbb{Q}[[t,u]], \tag*{(9)}

where the sum is over all forests TT whose component trees are mm-leafy, and n(T)n(T) and χm(T)\chi_m(T) are extended from trees to forests additively.

Write G=G(m)(t,u)G=G^{(m)}(t,u) and Gj≡Gj(m)(t):=[uj]G(m)(t,u)G_j\equiv G_j^{(m)}(t):=[u^j]G^{(m)}(t,u) for j≥0j\geq0, where G0=1G_0=1. For example, Gj(0)=1G_j(0)=1, which counts the forest of isolated jj leaves; G1G_1 is the generating series of the family {LTnm}n≥0\{\mathrm{LT}_n^m\}_{n\geq0}.

Proposition 3.7. The series GG and GjG_j, j=1,…,mj=1,\ldots,m satisfy the functional equation

tG2−(um(1−u)+t∑j=0mGjuj)G+um=0.(10)tG^2-\left(u^m(1-u)+t\sum_{j=0}^{m}G_j u^j\right)G+u^m=0. \tag*{(10)}

Note that (10) determines GG and G1,…,GmG_1,\ldots,G_m all at once. Solving for G1G_1 following the general methods in [6] leads to the closed formula (5) of ∣LTnm∣|\mathrm{LT}_n^m|.

Bijection to central tight gradings

While leafy trees are a new combinatorial object to our knowledge, they arose naturally in the study of tight gradings recently introduced by the authors to investigate rank-2 scattering diagrams [8]. In this section, we construct an explicit bijection between leafy trees and certain tight gradings. We begin by providing preliminaries on the relevant tight gradings, which are those corresponding to the central wall-coefficients of rank-2 skew-symmetric cluster scattering diagrams. Cluster algebras and their associated cluster scattering diagrams have an incredibly rich theory with numerous connections across mathematics and physics which we do not describe here; we instead refer the reader to [13, 14, 17] for further details.

Central tight gradings

We now give an explicit description of the tight gradings corresponding to the nthn^{\mathrm{th}} central wall coefficient of the cluster scattering diagram Dm+2,m+2\mathcal{D}_{m+2,m+2}. Tight gradings are defined much more generally in [7], and a tiling description is given in [9 Section 5].

For m,n∈Nm,n \in\mathbb{N}, let Pm,n\mathcal{P}_{m,n} be the NENE-path E(EN)(m+2)nE(EN)^{(m+2)n}. This path is sometimes referred to as the maximal Dyck path inside the rectangle of length (m+2)n+1(m+2)n+1 and height (m+2)n(m+2)n. Given a subset CC of steps (also known as edges) in an NENE-path P\mathcal{P}, we denote the set of east (resp. north) steps by CEC_{\mathrm{E}} (resp. CNC_{\mathrm{N}}) and the total number of steps by ∣C∣|C|.

A grading on an NENE-path is an assignment of a nonnegative integer to each edge of the path. Given a grading ω:P→Z≥0\omega:\mathcal{P}\to\mathbb{Z}_{\geq0} and a set of edges C⊂PC\subset\mathcal{P}, we set ω(C):=∑e∈Cω(e)\omega(C):=\sum_{e\in C}\omega(e). For edges u∈PEu\in\mathcal{P}_{\mathrm{E}} and v∈PNv\in\mathcal{P}_{\mathrm{N}}, let uv→\overrightarrow{uv} be the subpath proceeding east from the left endpoint of uu to the top endpoint of vv, continuing cyclically around P\mathcal{P} if necessary.

Definition 4.1. Fix an NENE-path P\mathcal{P} and a grading ω:P→Z≥0\omega:\mathcal{P}\to\mathbb{Z}_{\geq0}. For each edge ee in P\mathcal{P}, we define its shadow, denoted by sh⁡(e)\operatorname{sh}(e) or sh⁡(e;ω)\operatorname{sh}(e;\omega), as follows.

  • If e∈PEe\in\mathcal{P}_{\mathrm{E}}, then its shadow is ev→N\overrightarrow{ev}_{\mathrm{N}}, where v∈PNv\in\mathcal{P}_{\mathrm{N}} is chosen such that ∣ev→N∣=ω(ev→E)|\overrightarrow{ev}_{\mathrm{N}}|=\omega(\overrightarrow{ev}_{\mathrm{E}}) and ev→\overrightarrow{ev} has minimal length under this condition. If no such vv exists, let sh⁡(e)=PN\operatorname{sh}(e)=\mathcal{P}_{\mathrm{N}}.

  • If e∈PNe\in\mathcal{P}_{\mathrm{N}}, then its shadow is ue→E\overrightarrow{ue}_{\mathrm{E}}, where u∈PEu\in\mathcal{P}_{\mathrm{E}} is chosen such that ∣ue→E∣=ω(ue→N)|\overrightarrow{ue}_{\mathrm{E}}|=\omega(\overrightarrow{ue}_{\mathrm{N}}) and ue→\overrightarrow{ue} has minimal length under this condition. If no such uu exists, let sh⁡(e)=PE\operatorname{sh}(e)=\mathcal{P}_{\mathrm{E}}.

For C⊂PC\subset\mathcal{P}, let the shadow of CC be sh⁡(C)=⋃e∈Csh⁡(e)\operatorname{sh}(C)=\bigcup_{e\in C}\operatorname{sh}(e).

Definition 4.2. A grading ω\omega is compatible if for any edges u∈PEu\in\mathcal{P}_{\mathrm{E}} and v∈PNv\in\mathcal{P}_{\mathrm{N}}, either u∉sh⁡(v)u\notin\operatorname{sh}(v) or v∉sh⁡(u)v\notin\operatorname{sh}(u).

Compatible gradings (or compatible pairs) originated in the work of Lee-Li-Zelevinsky in their construction of the greedy basis [16].

Definition 4.3. A central tight grading of order nn is a compatible grading on Pm,n\mathcal{P}_{m,n} such that there are nn horizontal and nn vertical edges having value m+2m+2 in the grading, and all other edges have value 00 in the grading. We denote the set of all such tight gradings by TGnm\mathbf{TG}^{m}_{n}.

The following result is a special case of [8 Theorem 2.3].

Theorem 4.4. The central wall-function in Dm+2,m+2\mathcal{D}_{m+2,m+2} is given by

1+∑n≥1∣TGnm∣x(m+2)ny(m+2)n.1+\sum_{n\geq1}|\mathbf{TG}^{m}_{n}|x^{(m+2)n}y^{(m+2)n}.

Pictorially, we represent gradings by drawing a stack of height ∣sh⁡(u)∣|\operatorname{sh}(u)| consisting of (m+2)×1(m+2)\times1 blue rectangular tiles above each u∈PEu \in\mathcal{P}_{\mathrm{E}}, and a row of length ∣sh⁡(v)∣|\operatorname{sh}(v)| consisting of 1×(m+2)1\times(m+2) red rectangular tiles to the left of each v∈PNv \in\mathcal{P}_{\mathrm{N}}. Given a path P=UNEkNP=UNE^{k}N where UU is some NENE-word, we define the rotation operation ϱ\varrho which takes PP to be the NENE-path ϱ(P)=EkNUN\varrho(P)=E^{k}NUN. A rotation of PP is any path of the form ϱr(P)\varrho^{r}(P) for r≥0r\geq0. For the pictorial representation of a grading, it is sometimes useful to instead work on a rotation of the underlying NENE-path.

For each central tight grading ω\omega on Pn\mathcal{P}_n, there is a unique choice of f∈{0,1,…,(m+2)n−1}f\in\{0,1,\ldots,(m+2)n-1\} such that all tiles in the pictorial representation of the grading are contained in the rectangle bounding the rotated path ϱf\varrho^{f} [9], Remark 5.2. In this case, we say that ω\omega is ff-framed.

Remark 4.5. The compatibility condition on tight gradings implies that in this pictorial representation, the red and blue tiles do not overlap when drawn on any rotation of the underlying Dyck path.

Bijection between tight gradings and vertex-pointed leafy trees

A tree is vertex-pointed if it is equipped with a distinguished choice of vertex, denoted by vpointv_{\mathrm{point}}. Let LT^nm\widehat{\mathrm{LT}}{}^{m}_{n} denote the set of vertex-pointed mm-leafy trees on nn internal vertices and mn+1mn+1 leaves. Note that LT^nm≅LTnm×[mn+n+1]\widehat{\mathrm{LT}}{}^{m}_{n}\cong\mathrm{LT}^{m}_{n}\times[mn+n+1].

Fix T∈LT^nmT\in\widehat{\mathrm{LT}}{}^{m}_{n}. For each internal vertex vv, add an auxiliary vertex to be the rightmost child of vv. This yields a new tree aux⁡(T)\operatorname{aux}(T), which is an (m+1)(m+1)-leafy tree with mn+n+1mn+n+1 leaves and nn internal vertices. Note that the distinguished vertex vpointv_{\mathrm{point}} in aux⁡(T)\operatorname{aux}(T) can be any vertex except the auxiliary vertices.

Label all vertices of aux⁡(T)\operatorname{aux}(T) according to the left-to-right preorder, denoting the label of a vertex vv by λ(v)\lambda(v) and starting with label 0 at the root. We will work on the rotated NENE path ϱf(Pn)\varrho^{f}(\mathcal{P}_n), where f:=λ(vpoint)f:=\lambda(v_{\mathrm{point}}).

For each horizontal (resp. vertical) edge of ϱf(Pn)\varrho^{f}(\mathcal{P}_n), label it by uju_j (resp. vj−1v_{j-1}) where jj is the number of horizontal edges to its left in ϱf(Pn)\varrho^{f}(\mathcal{P}_n). Let ωE\omega_{\mathrm{E}} denote the set of horizontal edges with nonzero value in the grading ω\omega, and analogously for ωN\omega_{\mathrm{N}}.

Theorem 4.6. There is an explicit bijection

Γ^:LT^nm→TGnm(11)\widehat{\Gamma}: \widehat{\mathrm{LT}}_n^m \to\mathrm{TG}_n^m \tag*{(11)}

defined as follows. Given a vertex-pointed leafy tree (T,vpoint)∈LT^nm(T,v_{\mathrm{point}}) \in\widehat{\mathrm{LT}}_n^m with λ(v)\lambda(v) denoting the label of each vertex v∈aux⁡(T)v \in\operatorname{aux}(T), the corresponding tight grading ω:=Γ^((T,vpoint))\omega:= \widehat{\Gamma}((T,v_{\mathrm{point}})) on ϱλ(vpoint)(Pn)\varrho^{\lambda(v_{\mathrm{point}})}(P_n) is defined by

  • ωE={uj:λ(x)=j for some internal vertex x}\omega_{\mathrm{E}} = \{u_j : \lambda(x) = j\text{ for some internal vertex }x\}, and

  • ωN={vj:λ(a)=j for some auxiliary vertex a}\omega_{\mathrm{N}} = \{v_j : \lambda(a) = j\text{ for some auxiliary vertex }a\}.

Remark 4.8. The inverse map to Γ^\widehat{\Gamma} can be readily constructed as follows. Looking each horizontal edge ui∈ϱf(Pn)u_i \in\varrho^{f}(P_n), create a word w1w2⋯wnw_1w_2\cdots w_n on the 3-letter alphabet {(,),∙}\{(,),\bullet\} as follows:

wi={(if ω(ui)>0,)if ω(vi)>0, where vi is a north step immediately following ui,∙otherwise.w_i = \begin{cases} ( & \text{if }\omega(u_i)>0,\\ ) & \text{if }\omega(v_i)>0,\text{ where }v_i\text{ is a north step immediately following }u_i,\\ \bullet& \text{otherwise.} \end{cases}

The leafy tree TT is the plane tree associated to w1w2⋯wnw_1w_2\cdots w_n, and vpointv_{\mathrm{point}} is the vertex corresponding to the unique horizontal edge that is followed by another horizontal edge.

Bijection between 0-framed tight gradings and leafy trees

In order to construct a bijection involving leafy trees rather than the vertex-pointed leafy trees, we refine the tight gradings by their frame. Let TGn,fm\mathrm{TG}_{n,f}^m denote the subset of tight gradings in TGnm\mathrm{TG}_n^m that are ff-framed.

Note that the set of leafy trees in LTnm\mathrm{LT}_n^m is naturally identified with the subset of LT^nm\widehat{\mathrm{LT}}_n^m where the distinguished vertex is the root, in which case we say the tree is root-pointed. Equivalently, the root-pointed leafy trees in LT^nm\widehat{\mathrm{LT}}_n^m are those where λ(vpoint)=0\lambda(v_{\mathrm{point}})=0. Thus under the map Γ^\widehat{\Gamma}, the subset of root-pointed leafy trees in LT^nm\widehat{\mathrm{LT}}_n^m corresponds precisely to TGn,0m\mathrm{TG}_{n,0}^m.

These identifications together yield a direct bijection from leafy trees to 0-framed tight gradings.

Corollary 4.9. There is an explicit bijection

Γ:LTnm→TGn,0m(12)\Gamma: \mathrm{LT}_{n}^{m} \to\mathrm{TG}_{n,0}^{m} \tag*{(12)}

taking T∈LTnmT \in\mathrm{LT}_{n}^{m} with root v0v_0 to Γ^((T,v0))\widehat{\Gamma}((T,v_0)), as constructed in Theorem 4.6.

Example 4.11. Let TT be the 2-leafy tree at the left of Figure 7 (without a distinguished vertex). Then the corresponding tight grading Γ(T)\Gamma(T) is depicted on the left of Figure 6.

A 0-framed central tight grading and a 7-framed central tight grading

Figure 6. A 0-framed central tight grading (left) and a 7-framed central tight grading (right) where m=1m=1 and n=3n=3. Note that edges of the underlying NENE-path that are directly below a blue tile or to the right of a red tile are precisely the edges with value m+2=3m+2=3 in the tight grading, and all other edges take value 0.

A vertex-pointed leafy tree, its auxiliary tree, and the corresponding tight grading

Figure 7. For a vertex-pointed 1-leafy tree TT (left), we depict the tree aux⁡(T)\operatorname{aux}(T) (middle) where vpointv_{\mathrm{point}} is circled in each. On the right, we have the corresponding tight grading ω\omega, where ωE={u0,u1,u3}\omega_{\mathrm{E}} = \{u_0,u_1,u_3\} and ωN={v6,v8,v9}\omega_{\mathrm{N}} = \{v_6,v_8,v_9\}. Note that the tight grading is 7-framed, where 7 is the label of the distinguished vertex in aux⁡(T)\operatorname{aux}(T).

Acknowledgements

The authors are grateful to Hyun Kyu Kim and Korea Institute for Advanced Study, where several key ideas were developed when AB and LM visited KL in June 2026. The authors also thank François Bergeron, Mireille Bousquet-Mélou, Frédéric Chapoton, Éric Fusy, Vincent Pilaud, Markus Reineke, and Hugh Thomas for helpful discussions and comments. AB was partially supported by NSF grant DMS-2503411.

AI disclosure. The authors formulated the definition of leafy trees, inspired by their study of tight gradings, without AI assistance. After this, the only notable mathematical AI contribution came from querying GPT-5.6 in July 2026 for ways to associate a Dyck path to a 1-leafy tree. Among the suggestions were the constructions we later generalized to define the bijection Φ\Phi. The proofs (to be presented in a later full version of this article) were discovered without AI assistance. AI was used elsewhere in reviewing exposition, performing literature searches, doing numerical experiments, and generating figures.

References

Amanda Burcroff, Department of Mathematics, MIT, Cambridge, MA 02139, USA

Email address: [email protected]

Kyungyong Lee, Department of Mathematics, University of Alabama, Tuscaloosa, AL 35487, USA and Korea Institute for Advanced Study, Seoul 02455, Republic of Korea

Email address: [email protected] & [email protected]

Lang Mou, Department of Mathematics, University of California Davis, Davis, CA 95616, USA

Email address: [email protected]

Figure 1 (PDF p. 13). The 2-Tamari lattice T32\mathcal{T}_3^2. The remaining 30 intervals, grouped by length. ℓ([P,Q])=max⁡{k:P=P0≺P1≺⋯≺Pk=Q}\ell([P,Q])=\max\{k:P=P_0\prec P_1\prec\cdots\prec P_k=Q\}. The intervals are grouped under ℓ=2\ell=2, ℓ=3\ell=3, ℓ=4\ell=4, ℓ=5\ell=5, and ℓ=6\ell=6.

References

  1. [1]Alexandre Astruc, Frédéric Chapoton, Karen Martinez, and Markus Reineke, Motives of central slope Kronecker moduli, Épijournal Géom. Algébrique 9 (2025), Art. 19, 15.DOI
  2. [2]François Bergeron and Louis-François Préville-Ratelle, Higher trivariate diagonal harmonics via generalized Tamari posets, J. Comb. 3 (2012), no. 3, 317–341.arxiv.org/abs/1105.3738
  3. [3]Olivier Bernardi and Nicolas Bonichon, Intervals in Catalan lattices and realizers of triangulations, J. Combin. Theory Ser. A 116 (2009), no. 1, 55–75.DOI
  4. [4]Mireille Bousquet-Mélou, Rational and algebraic series in combinatorial enumeration, International Congress of Mathematicians. Vol. III, Eur. Math. Soc., Zürich, 2006, pp. 789–826.
  5. [5]Mireille Bousquet-Mélou, Éric Fusy, and Louis-François Préville-Ratelle, The number of intervals in the m-Tamari lattices, Electron. J. Combin. 18 (2011), no. 2, Paper 31, 26.
  6. [6]Mireille Bousquet-Mélou and Arnaud Jehanne, Polynomial equations with one catalytic variable, algebraic series and map enumeration, J. Combin. Theory Ser. B 96 (2006), no. 5, 623–672.
  7. [7]Amanda Burcroff, Kyungyong Lee, and Lang Mou, Positivity of generalized cluster scattering diagrams, arXiv preprint 2503.03719, 2025.arxiv.org/abs/2503.03719
  8. [8]————, Scattering diagrams, tight gradings, and generalized positivity, Proc. Natl. Acad. Sci. USA 122 (2025), no. 18, e2422893122.DOI
  9. [9]Amanda Burcroff, Kyungyong Lee, Lang Mou, Gregg Musiker, and Markus Reineke, Cluster scattering diagrams via quiver moduli and tight gradings, J. Lond. Math. Soc. (2) 114 (2026), no. 2, Paper No. e70642.arxiv.org/abs/2511.14672
  10. [10]F. Chapoton, Sur le nombre d’intervalles dans les treillis de Tamari, Sém. Lothar. Combin. 55 (2005/07), Art. B55f, 18.
  11. [11]Grégory Châtel and Viviane Pons, Counting smaller elements in the Tamari and m-Tamari lattices, J. Combin. Theory Ser. A 134 (2015), 58–97.
  12. [12]Wenjie Fang, Éric Fusy, and Philippe Nadeau, Tamari intervals and blossoming trees, Comb. Theory 5 (2025), no. 1, Paper No. 4, 41.DOI
  13. [13]Sergey Fomin, Lauren Williams, and Andrei Zelevinsky, Introduction to cluster algebras. Chapters 1–3, arXiv preprint 1608.05735, 2016.arxiv.org/abs/1608.05735
  14. [14]Mark Gross, Paul Hacking, Sean Keel, and Maxim Kontsevich, Canonical bases for cluster algebras, J. Amer. Math. Soc. 31 (2018), no. 2, 497–608.
  15. [15]Mark Gross and Rahul Pandharipande, Quivers, curves, and the tropical vertex, Port. Math. 67 (2010), no. 2, 211–259.arxiv.org/abs/0909.5153
  16. [16]Kyungyong Lee, Li Li, and Andrei Zelevinsky, Greedy elements in rank 2 cluster algebras, Selecta Math. (N.S.) 20 (2014), no. 1, 57–82.
  17. [17]Tomoki Nakanishi, Cluster algebras and scattering diagrams, MSJ Memoirs, vol. 41, Mathematical Society of Japan, Tokyo, 2023.DOI
  18. [18]Viviane Pons, The rise-contact involution on Tamari intervals, Electron. J. Combin. 26 (2019), no. 2, Paper No. 2.32, 53.arxiv.org/abs/1802.08335
  19. [19]Markus Reineke, Cohomology of quiver moduli, functional equations, and integrality of Donaldson-Thomas type invariants, Compos. Math. 147 (2011), no. 3, 943–964.DOI
  20. [20]Markus Reineke and Thorsten Weist, Refined GW/Kronecker correspondence, Math. Ann. 355 (2013), no. 1, 17–56.
  21. [21]Thorsten Weist, Localization in quiver moduli spaces, Represent. Theory 17 (2013), 382–425.arxiv.org/abs/0903.5442

Paper details

vlnc:W_wb4fu983
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 12

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

GPT-5.6, AI

View statement
Looking for related papers…
Contents