This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

Flips on homologous orientations of surface graphs with prescribed forbidden facial circuits

Weijuan Zhang1, Jianguo Qian2111Corresponding author, email address: [email protected]
1School of Mathematical Sciences, Xinjiang Normal University
Urumqi, Xinjiang 830054, P.R.China
2School of Mathematical Sciences, Xiamen University
Xiamen, Fujian 361005, P.R.China
Abstract

Let GG be a graph embedded on an orientable surface. Given a class 𝒞{\cal C} of facial circuits of GG as a forbidden class, we give a sufficient-necessary condition for that an α\alpha-orientation (orientation with prescribed out-degrees) of GG can be transformed into another by a sequence of flips on non-forbidden circuits and further give an explicit formula for the minimum number of such flips. We also consider the connection among all α\alpha-orientations by defining a directed graph 𝐃(𝒞){\bf D}({\cal C}), namely the 𝒞{\cal C}-forbidden flip graph. We show that if 𝒞{\cal C}\not=\emptyset, then 𝐃(𝒞){\bf D}({\cal C}) has exactly |O(G,𝒞)||O(G,{\cal C})| components, each of which is the cover graph of a distributive lattice, where |O(G,𝒞)||O(G,{\cal C})| is the number of the α\alpha-orientations that has no counterclockwise facial circuit other than that in 𝒞{\cal C}. If 𝒞={\cal C}=\emptyset, then every component of 𝐃(𝒞){\bf D}({\cal C}) is strongly connected. This generalizes the corresponding results of Felsner and Propp for the case that 𝒞{\cal C} consists of a single facial circuit.

Keywords: surface graph, orientation, flip, homology, forbidden facial circuit

1 Introduction

Graph orientation with certain degree-constraints on vertices is a natural focus in graph theory and has close connection with many combinatorial structures in graphs, such as the spanning trees [2], primal-dual orientations [1], bipolar orientations [4], transversal structures [5], Schnyder woods [6], bipartite perfect matchings (or more generally, bipartite ff-factors) [8, 10, 12] and cc-orientations of the dual of a plane graph [9, 12]. In general, all these constraints could be encoded in terms of α\alpha-orientations [2], that is, the orientations with prescribed out-degrees.

To deal with the relation among orientations, circuit (or cycle) reversal has been shown as a useful method since it preserves the out-degree of each vertex and the connectivity of the orientations. Various types of cycle reversals were introduced subject to certain requirements. Circuit reversal for the orientations of plane graphs or, more generally, surface graphs (graphs embedded on an orientable surface) received particular attention [2, 6, 9, 10, 11]. For example, Nakamoto [11] considered the 3-cycle reversal to deal with those orientations in plane triangulations where each vertex on the outer facial cycle has out-degree 1 while each of the other vertices has out-degree 3. In [13], Zhang et al. introduced the Z-transformation to study the connection among perfect matchings in hexagonal systems, which was later extended to general plane bipartite graphs [14].

A natural considering of circuit reversal for orientations of a surface graph is to reverse a directed facial circuit (the circuit that bounds a face). However, an orientation of a surface graph does not always have such a directed facial circuit, even if it has many directed circuits. In [2], Felsner introduced a type of circuit reversals, namely the flip, defined on the essential cycles in plane graphs which reverses a counterclockwise essential cycle to be clockwise. In a strongly connected α\alpha-orientation of a plane graph, every essential cycle is exactly an inner facial circuit. In this sense, the notion of essential cycle is a very nice generalization of that of facial circuit. Further, Felsner proved that any α\alpha-orientation of a plane graph can be transformed into a particular α\alpha-orientation by a flip sequence, and the set of all α\alpha-orientations of the graph carries a distributive lattice by flips.

The notion of flip has been also extended to general surface graphs [6, 12] and oriented matroids [9]. Moreover, Propp proved the following result (where the notion of homology will be given in the next section).

Theorem 1.1.

[12] The set of all α\alpha-orientations of a surface graph GG in the same homology class carries a distributive lattice by a sequence of flips taking on the facial circuits of GG with a prescribed forbidden facial circuit.

We note that the above theorem is an extension of the one given by Felsner for plane graphs since a plane graph can be regarded as a sphere graph where the outer facial circuit is treated as the forbidden circuit.

In this paper, in stead of a prescribed forbidden facial circuit, we consider the flips on surface graph with a class of prescribed forbidden facial circuits 𝒞{\cal C}, and call such flips the 𝒞{\cal C}-forbidden flips. In the third section, we give a sufficient and necessary condition for that an α\alpha-orientation can be transformed into another by a sequence of 𝒞{\cal C}-forbidden flips and, further, give an explicit formula for the minimum number of the flips needed in such a sequence. In particular, if 𝒞{\cal C} is empty then an α\alpha-orientation can be transformed into another by flips if and only if they are homologous. In our study, the idea of potential function [2], or depth function [11], plays an important role.

In the last section, we focus on the relationship among all α\alpha-orientations in a surface graph GG. To this end, given a class 𝒞{\cal C} of forbidden facial circuits, we define a directed graph, namely the 𝒞{\cal C}-forbidden flip graph, whose vertex set is the class of all α\alpha-orientations of GG and two orientations DD and DD^{\prime} are joined by an edge with direction from DD^{\prime} to DD provided DD can be transformed from DD^{\prime} by a 𝒞{\cal C}-forbidden flip. We apply the technique of U-coloring and L-coloring introduced by Felsner and Knauer in [3], and show that the 𝒞{\cal C}-forbidden flip graph admits both a U- and L-coloring for any class 𝒞{\cal C}. This yields a generalization of Theorem 1.1. More specifically, if 𝒞{\cal C} is not empty, then the flip graph has exactly |O(G,𝒞)||O(G,{\cal C})| components, each of which is the cover graph of a distributive lattice, where |O(G,𝒞)||O(G,{\cal C})| is the number of the α\alpha-orientations that has no counterclockwise facial circuit other than that in 𝒞{\cal C}. And if 𝒞{\cal C} is empty, then every non-trivial component of the flip graph is strongly connected, and hence cyclic.

2 Preliminaries

In this section we introduce some elementary concepts involving graphs and graph embeddings on surfaces. For a graph GG (directed or undirected), we denote by V(G)V(G) and E(G)E(G) the vertex set and edge set of GG, respectively. A closed walk in a graph is a sequence v1v2vkv_{1}v_{2}\dotsc v_{k} such that v1v_{1} and vkv_{k} are adjacent and, for every i{1,2,,k1}i\in\{1,2,\dotsc,k-1\}, viv_{i} and vi+1v_{i+1} are adjacent. A closed walk is called a cycle if it is edge disjoint and a circuit if it is vertex disjoint. The directed circuit and directed cycle are defined analogously. An orientation of a graph GG is an assignment of a direction to each edge of GG. Given a graph GG with nn vertices v1,v2,,vnv_{1},v_{2},\dotsc,v_{n} and an out-degree sequence α=(α(v1),α(v2),,α(vn))\alpha=(\alpha(v_{1}),\alpha(v_{2}),\dotsc,\alpha(v_{n})), an orientation DD of GG is called an α\alpha-orientation [2] if dD+(vi)=α(vi)d^{+}_{D}(v_{i})=\alpha(v_{i}) for all viV(G)v_{i}\in V(G), where dD+(vi)d^{+}_{D}(v_{i}) is the out-degree of viv_{i} in DD. It is well known that if an α\alpha-orientation of GG is strongly connected, then every α\alpha-orientation of GG is strongly connected. In this case, we call α\alpha strongly connected. It is also known that each component obtained from an α\alpha-orientation by deleting all those edges that have the same direction in every α\alpha-orientation of GG (i.e., the rigid edges [2, 6]) is strongly connected [9]. So in this paper, we always assume that α\alpha is strongly connected and, hence, GG is 2-edge connected.

A surface graph is a graph GG embedded on an orientable surface Σ\Sigma without boundary. The connected components of ΣG\Sigma\setminus G, when viewed as subsets of Σ\Sigma, are called the faces of GG. A closed walk that bounds a face is called a facial walk. A surface graph is called a 2-cell embedding [7] or map [6] if all its faces are homeomorphic to open disks. A cycle CC on a surface Σ\Sigma is separating if ΣC\Sigma\setminus C is disconnected and is contractible [7] if CC can be continuously transformed into a single point. We note that if a cycle is contractible, then it is separating and, moreover, one component of ΣC\Sigma\setminus C is homeomorphic to an open disc. In this paper, except where otherwise stated, we always assume that all the surface graphs are 2-cell embedded and every edge is on the boundary of two distinct faces. We note that, in such an embedding, a facial walk is exactly a facial circuit and, therefore, every edge is shared by two distinct facial circuits. Thus, our embedding is stronger than 2-cell embedding, but weaker than the one in which each facial circuit is contractible.

A directed facial circuit FF is called counterclockwise (resp., clockwise), or ccw (resp., cw) for short, if when we trace along with the orientation of FF, the face bounded by FF lies to the left (resp., right) side of FF. A flip taking on a ccw facial circuit FF is the reorientation of FF that reverses FF from ccw to cw [2]. The notion of flip was introduced initially for the essential circuits. As mentioned earlier, in a strongly connected α\alpha-orientation of GG, every essential circuit is a facial circuit [9].

Let GG be a surface graph with edge set EE of mm edges and let D0D_{0} be a fixed orientation of GG. Consider the mm-dimensional edge space |E|\mathbb{Z}^{|E|} of GG. It is clear that any oriented subgraph DD of GG can be represented as an mm-dimensional vector ϕ(D)|E|\phi(D)\in\mathbb{Z}^{|E|} whose coordinate indexed by an edge ee is defined by

ϕ(D)e={ 1,if e has the same orientation in D0 and D,1,if e has the opposite orientation in D0 and D, 0,if e is not in D.\phi(D)_{e}=\begin{cases}\ \ \ 1,&\text{if\ $e$\ has\ the\ same\ orientation\ in\ $D_{0}$\ and\ $D$},\\ \ -1,&\text{if\ $e$\ has\ the\ opposite\ orientation\ in\ $D_{0}$\ and\ $D$},\\ \ \ \ 0,&\text{if\ $e$\ is\ not\ in\ $D$}.\end{cases}

Moreover, all oriented even subgraphs (the in-degree of each vertex equals its out-degree) of GG form an (mn+1)(m-n+1)-dimensional subspace of |E|\mathbb{Z}^{|E|}, called the (directed) cycle space of GG, where nn is the number of vertices in GG.

For a simple graph GG, a basis of the cycle space can be created by the elementary cycles associated with a given spanning tree of GG. For a graph (not necessarily simple) embedded on a surface, it would be more convenient to create a basis of the cycle space by the facial circuits in {F0}{\cal F}\setminus\{F_{0}\} and a so-called basis for the homology, where and herein after, {\cal F} denotes the set of all facial circuits of GG and F0F_{0} is an arbitrary facial circuit in {\cal F}. A common approach to create a basis for the homology is to choose a spanning tree TT in GG and a spanning tree TT^{*} in the dual graph GG^{*} of GG that contains no edge dual to TT. By Euler’s formula, there are exactly 2g2g edges in GG that are neither in TT nor dual to the edges in TT^{*}, where gg is the genus of the surface. Each of these 2g2g edges forms a unique circuit with TT and all these 2g2g circuits form a basis for the homology [6], denoted by {\cal H}. In a word, any oriented even subgraph DD of GG can be uniquely represented as a linear combination of the facial circuits in {F0}{\cal F}\setminus\{F_{0}\} and circuits in {\cal H}, i.e.,

ϕ(D)=F{F0}λFϕ(F)+HμHϕ(H).\phi(D)=\sum_{F\in{\cal F}\setminus\{F_{0}\}}\lambda_{F}\phi(F)+\sum_{H\in{\cal H}}\mu_{H}\phi(H). (1)

In the following we always choose the orientation of each facial circuit in {\cal F} to be ccw. Let 𝔽\mathbb{F} be the subspace of |E|\mathbb{Z}^{|E|} generated by {\cal F}. An oriented even subgraph DD is called null-homologous or 0-homologous [6] if ϕ(D)𝔽\phi(D)\in\mathbb{F}, i.e.,

ϕ(D)=FλFϕ(F).\phi(D)=\sum_{F\in{\cal F}}\lambda_{F}\phi(F). (2)

For two α\alpha-orientations DD and DD^{\prime}, we denote by DDD^{\prime}-D the directed subgraph obtained from DD^{\prime} by removing those edges that has the same direction as that in DD. It is clear that DDD^{\prime}-D is oriented even graph. Further, if ϕ(DD)𝔽\phi(D^{\prime}-D)\in\mathbb{F} then DD and DD^{\prime} are called homologous [6]. Equivalently, DD and DD^{\prime} are homologous if and only if their corresponding coefficients μH\mu_{H} in (1) are the same for every HH\in{\cal H}. It is clear that homology is an equivalence relation on α\alpha-orientations.

Example 1. Let GG be the Cartesian product graph of two circuits of length 2 embedded on torus and α=(2,2,2,2)\alpha=(2,2,2,2). For convenience, we draw GG in the form as indicated in Figure 1 (left), where the left side and right side (resp., lower side and upper side) are identified. By the method we introduced earlier, we can see that ={HM,HP}{\cal H}=\{H_{\rm M},H_{\rm P}\} is a basis for the homology, where HMH_{\rm M} and HPH_{\rm P} are the two directed circuits as illustrated in Figure 1 (right).

[Uncaptioned image]

Figure 1. The edges of GG are represented in thick lines; the four faces are labeled by F1,F2,F3,F4F_{1},F_{2},F_{3},F_{4}.

We note that, since every vertex has out-degree 2, the directions of the four edges on F1F_{1} are determined uniquely by that on F4F_{4}, unless F4F_{4} is a directed circuit (in this case F1F_{1} must be directed and, therefore, has exactly two choices). This means that GG has totally 18 different (2,2,2,2)(2,2,2,2)-orientations D1,D2,D_{1},D_{2}, ,D18\dotsc,D_{18}, as illustrated in Figure 2. Since D1D2=HMD_{1}-D_{2}=H_{\rm M}, we have ϕ(D1D2)=ϕ(HM)\phi(D_{1}-D_{2})=\phi(H_{\rm M}), meaning that D1D_{1} and D2D_{2} are not homologous. Further, since D13D14=D13D_{13}-D_{14}=D_{13} and ϕ(D13)=ϕ(F1)+ϕ(F4)\phi(D_{13})=\phi(F_{1})+\phi(F_{4}), D13D_{13} and D14D_{14} are homologous. In general, it can be seen that any two orientations DiD_{i} and DjD_{j} are homologous if and only if i=ji=j or i,j{13,14,,18}i,j\in\{13,14,\dotsc,18\}. In other words, the (2,2,2,2)(2,2,2,2)-orientations of GG form 13 homologous classes: {D1},{D2},,{D12}\{D_{1}\},\{D_{2}\},\dotsc,\{D_{12}\} and {D13,D14,,D18}\{D_{13},D_{14},\dotsc,D_{18}\}.

[Uncaptioned image]

Figure 2.

3 Transform an α\alpha-orientation into another by flips on non-forbidden facial circuits

Proposition 3.1.

Let GG be a surface graph and, for each FF\in{\cal F}, let λF\lambda_{F} be an integer associated with FF. Then for any integer kk,

FλFϕ(F)=F(λF+k)ϕ(F).\sum_{F\in\mathcal{F}}\lambda_{F}\phi(F)=\sum_{F\in\mathcal{F}}(\lambda_{F}+k)\phi(F). (3)

Conversely, for each FF\in{\cal F}, if ηF\eta_{F} is an integer associated with FF such that

FηFϕ(F)=FλFϕ(F),\sum_{F\in\mathcal{F}}\eta_{F}\phi(F)=\sum_{F\in\mathcal{F}}\lambda_{F}\phi(F), (4)

then there exists an integer kk such that ηF=λF+k\eta_{F}=\lambda_{F}+k for every FF\in\mathcal{F}.

Proof.

Consider an edge ee of GG. By our assumption, ee is shared by two distinct facial circuits, say, FF and FF^{\prime}. Recall that every facial circuit in {\cal F}, when viewed as a basis circuit, is ccw. Therefore, ee has opposite orientations in FF and FF^{\prime}. Thus, kϕ(F)e+kϕ(F)e=0k\phi(F)_{e}+k\phi(F^{\prime})_{e}=0 and ϕ(K)e=0\phi(K)_{e}=0 for any KK\in{\cal F} other than FF and FF^{\prime}. Thus, (3) follows directly.

Conversely, let ee be an arbitrary edge and let FF and FF^{\prime} be the two facial circuits that share ee. Then by (4),

μFϕ(F)e+ηFϕ(F)e=λFϕ(F)e+λFϕ(F)e.\mu_{F}\phi(F)_{e}+\eta_{F^{\prime}}\phi(F^{\prime})_{e}=\lambda_{F}\phi(F)_{e}+\lambda_{F^{\prime}}\phi(F^{\prime})_{e}. (5)

Further, we assume that ηF=λF+k\eta_{F}=\lambda_{F}+k and ηF=λF+k\eta_{F^{\prime}}=\lambda_{F^{\prime}}+k^{\prime}. Then,

ηFϕ(F)e+ηFϕ(F)e=(λF+k)ϕ(F)e+(λF+k)ϕ(F)e\eta_{F}\phi(F)_{e}+\eta_{F^{\prime}}\phi(F^{\prime})_{e}=(\lambda_{F}+k)\phi(F)_{e}+(\lambda_{F^{\prime}}+k^{\prime})\phi(F^{\prime})_{e}
=(λFϕ(F)e+λFϕ(F)e)+(kϕ(F)e+kϕ(F)e.=(\lambda_{F}\phi(F)_{e}+\lambda_{F^{\prime}}\phi(F^{\prime})_{e})+(k\phi(F)_{e}+k^{\prime}\phi(F^{\prime})_{e}.

So by (5), we have kϕ(F)e+kϕ(F)e=0k\phi(F)_{e}+k^{\prime}\phi(F^{\prime})_{e}=0. Recall that both FF and FF^{\prime} are ccw. Therefore, ϕ(F)e=ϕ(F)e\phi(F)_{e}=-\phi(F^{\prime})_{e} and, hence, k=kk=k^{\prime}. Further, notice that the dual graph of GG is connected, meaning that ηF=λF+k\eta_{F}=\lambda_{F}+k for all FF\in\mathcal{F}. ∎

Let DD and DD^{\prime} be two homologous α\alpha-orientations of a surface graph GG. Following the idea of potential function [2] (or depth function [11]), we introduce an intuitive representation of ϕ(DD)\phi(D^{\prime}-D) as follows: define

zDD:𝐙z_{D^{\prime}-D}:{\cal F}\rightarrow\mathbf{Z}

to be the function which assigns an integer to each facial circuit FF of GG according to the following rule, see Figure 1:
1. zDD(F0)=0z_{D^{\prime}-D}(F_{0})=0;
2. if FF and FF^{\prime} share a common edge ee then

zDD(F)={zDD(F)+1,if eDD and F lies to the left of e,zDD(F)1,if eDD and F lies to the left of e,zDD(F),if eDD,z_{D^{\prime}-D}(F)=\begin{cases}\ z_{D^{\prime}-D}(F^{\prime})+1,&\text{if\ $e\in D^{\prime}-D$\ and\ $F$\ lies\ to\ the\ left\ of\ $e$},\\ \ z_{D^{\prime}-D}(F^{\prime})-1,&\text{if\ $e\in D^{\prime}-D$\ and\ $F^{\prime}$\ lies\ to\ the\ left\ of\ $e$},\\ \ z_{D^{\prime}-D}(F^{\prime}),&\text{if\ $e\notin D^{\prime}-D$},\\ \end{cases} (6)

where, by ‘FF lies to the left of ee’ we mean that FF lies to the left side of ee when we trace along with the direction of ee.

[Uncaptioned image]

Figure 3. The edges in DDD^{\prime}-D are represented in thick lines and edges not in DDD^{\prime}-D are represented in dotted lines.

Proposition 3.2.

For any two homologous α\alpha-orientations DD and DD^{\prime}, zDD(F)z_{D^{\prime}-D}(F) is well defined and

ϕ(DD)=FzDD(F)ϕ(F).\phi(D^{\prime}-D)=\sum_{F\in\mathcal{F}}z_{D^{\prime}-D}(F)\phi(F).
Proof.

Since DD and DD^{\prime} are homologous, we have ϕ(DD)=FλFϕ(F)\phi(D^{\prime}-D)=\sum_{F\in{\cal F}}\lambda_{F}\phi(F). Let ee be an arbitrary edge of GG, and let FF and FF^{\prime} be the two facial circuits that share ee. Then ϕ(DD)e=λFϕ(F)e+λFϕ(F)e\phi(D^{\prime}-D)_{e}=\lambda_{F}\phi(F)_{e}+\lambda_{F^{\prime}}\phi(F^{\prime})_{e}.

We first assume that the direction of ee in DDD^{\prime}-D coincides with that in our initially fixed orientation D0D_{0}, i.e., ϕ(DD)e=ϕ(D0)e=1\phi(D^{\prime}-D)_{e}=\phi(D_{0})_{e}=1. In this case, we have ϕ(F)e=1\phi(F)_{e}=1 and ϕ(F)e=1\phi(F^{\prime})_{e}=-1 if FF lies to the left of ee, and ϕ(F)e=1\phi(F)_{e}=-1 and ϕ(F)e=1\phi(F^{\prime})_{e}=1 if FF lies to the right of ee. Therefore,

λF={λF+1,if eDD and F lies to the left of e,λF1,if eDD and F lies to the left of e,λF,if eDD.\lambda_{F}=\begin{cases}\ \lambda_{F^{\prime}}+1,&\text{if\ $e\in D^{\prime}-D$\ and\ $F$\ lies\ to\ the\ left\ of\ $e$},\\ \ \lambda_{F^{\prime}}-1,&\text{if\ $e\in D^{\prime}-D$\ and\ $F^{\prime}$\ lies\ to\ the\ left\ of\ $e$},\\ \ \lambda_{F^{\prime}},&\text{if\ $e\notin D^{\prime}-D$}.\\ \end{cases} (7)

We now assume that ϕ(DD)e=ϕ(D0)e=1\phi(D^{\prime}-D)_{e}=-\phi(D_{0})_{e}=-1. In this case we have ϕ(F)e=1\phi(F)_{e}=-1 and ϕ(F)e=1\phi(F^{\prime})_{e}=1 if FF lies to the left of ee, and ϕ(F)e=1\phi(F)_{e}=1 and ϕ(F)e=1\phi(F^{\prime})_{e}=-1 if FF lies to the right of ee. This means that (7) still holds.

Combining (7) with (6), we have

λFλF=zDD(F)zDD(F)\lambda_{F}-\lambda_{F^{\prime}}=z_{D^{\prime}-D}(F)-z_{D^{\prime}-D}(F^{\prime}) (8)

and, hence, λFzDD(F)=λFzDD(F)\lambda_{F}-z_{D^{\prime}-D}(F)=\lambda_{F^{\prime}}-z_{D^{\prime}-D}(F^{\prime}). Since the dual graph of GG is connected, we have

λFzDD(F)=λKzDD(K)\lambda_{F}-z_{D^{\prime}-D}(F)=\lambda_{K}-z_{D^{\prime}-D}(K) (9)

for any KK\in{\cal F}. In particular, λFzDD(F)=λF0zDD(F0)=λF0\lambda_{F}-z_{D^{\prime}-D}(F)=\lambda_{F_{0}}-z_{D^{\prime}-D}(F_{0})=\lambda_{F_{0}}, i.e., zDD(F)=λFλF0z_{D^{\prime}-D}(F)=\lambda_{F}-\lambda_{F_{0}}. Therefore, zDD(F)z_{D^{\prime}-D}(F) is well-defined. Further, the proposition follows directly from (9) and Proposition 3.1. ∎

Lemma 3.3.

[6] Let DD and DD^{\prime} be two α\alpha-orientations of a surface graph and let F0F_{0} be an arbitrary facial circuit. If

ϕ(DD)=F{F0}λFϕ(F)\phi(D^{\prime}-D)=\sum_{F\in\mathcal{F}\setminus\{F_{0}\}}\lambda_{F}\phi(F)

and λF0\lambda_{F}\geq 0 for all F{F0}F\in\mathcal{F}\setminus\{F_{0}\}, then DD can be transformed from DD^{\prime} by a sequence of {F0}\{F_{0}\}-forbidden flips.

Let

zmin=min{zDD(F):F(G)}z_{\min}=\min\{z_{D^{\prime}-D}(F):F\in{\cal F}(G)\}

and let FminF_{\min} be a facial circuit for which zDD(Fmin)=zminz_{D^{\prime}-D}(F_{\min})=z_{\min}.

Theorem 3.4.

Given a forbidden facial circuit class 𝒞{\cal C}, an α\alpha-orientation DD of a surface graph can be transformed from DD^{\prime} by a sequence of 𝒞{\cal C}-forbidden flips if and only if DD and DD^{\prime} are homologous and zDD(F)=zminz_{D^{\prime}-D}(F)=z_{\min} for every F𝒞F\in{\cal C}. Moreover, the minimum number of flips needed to transform DD^{\prime} into DD equals

d(D,D)=F(zDD(F)zmin).d(D^{\prime},D)=\sum\limits_{F\in{\cal F}}(z_{D^{\prime}-D}(F)-z_{\min}). (10)
Proof.

Assume firstly that DD can be transformed from DD^{\prime} by a sequence 𝒮{\cal S} of 𝒞{\cal C}-forbidden flips.

For a facial circuit FF\in{\cal F}, consider the number t(F)t(F) of the flips in 𝒮{\cal S} taking on FF. Let ee be an edge on FF and let FF^{\prime} be the facial circuit which shares ee with FF. If eDDe\in D^{\prime}-D, then the orientation of ee changes after 𝒮{\cal S} being taken. Moreover, a flip to change the orientation of ee must take on the facial circuit which lies to the left side of ee. This implies that t(F)=t(F)+1t(F)=t(F^{\prime})+1 if FF lies to the left side of ee or t(F)=t(F)1t(F)=t(F^{\prime})-1 if FF^{\prime} lies to the left side of ee. If eDDe\notin D^{\prime}-D, then the orientation of ee does not change after 𝒮{\cal S} being taken. This means that, if a flip in 𝒮{\cal S} takes on FF then there must be another flip that takes on FF^{\prime} to keep the orientation of ee invariant and, hence, t(F)=t(F)t(F)=t(F^{\prime}). As a result, we have the following relation:

t(F)={t(F)+1,if eDD and F lies to the left  side of e,t(F)1,if eDD and F lies to the left  side of e,t(F),if eDD.t(F)=\begin{cases}\ t(F^{\prime})+1,&\text{if\ $e\in D^{\prime}-D$\ and\ $F$\ lies\ to\ the\ left\ side\ of\ $e$},\\ \ t(F^{\prime})-1,&\text{if\ $e\in D^{\prime}-D$\ and\ $F^{\prime}$\ lies\ to\ the\ left\ side\ of\ $e$},\\ \ t(F^{\prime}),&\text{if\ $e\notin D^{\prime}-D$}.\end{cases} (11)

Combining (11) with (6), we have t(F)t(F)=zDD(F)zDD(F)t(F)-t(F^{\prime})=z_{D^{\prime}-D}(F)-z_{D^{\prime}-D}(F^{\prime}), i.e., t(F)zDD(F)=t(F)zDD(F)t(F)-z_{D^{\prime}-D}(F)=t(F^{\prime})-z_{D^{\prime}-D}(F^{\prime}). Since the dual graph of GG is connected, we have

t(F)zDD(F)=t(K)zDD(K)t(F)-z_{D^{\prime}-D}(F)=t(K)-z_{D^{\prime}-D}(K) (12)

for any KK\in{\cal F}. So by Proposition 3.1 and Proposition 3.2,

ϕ(DD)=Ft(F)ϕ(F).\phi(D^{\prime}-D)=\sum_{F\in\mathcal{F}}t(F)\phi(F).

Hence, DD and DD^{\prime} are homologous.

Further, from (12) we can also see that t(K)t(K) attains the minimum value if and only if zDD(K)z_{D^{\prime}-D}(K) does, i.e., zDD(K)=zminz_{D^{\prime}-D}(K)=z_{\min} and K=FminK=F_{\min}. On the other hand, notice that t(F)0t(F)\geq 0 for every FF\in{\cal F} and the minimum value 0 of t(F)t(F) is attained by every F𝒞F\in{\cal C} since the facial circuits in 𝒞{\cal C} do not involve any flip in 𝒮{\cal S}. Hence, t(Fmin)=t(F)=0t(F_{\min})=t(F)=0 for every F𝒞F\in{\cal C}. Therefore,

zDD(F)=zDD(Fmin)=zminz_{D^{\prime}-D}(F)=z_{D^{\prime}-D}(F_{\min})=z_{\min} (13)

for every F𝒞F\in{\cal C}.

Conversely, assume that DD and DD^{\prime} are homologous and zDD(F)=zminz_{D^{\prime}-D}(F)=z_{\min} for any F𝒞F\in{\cal C}. Then by Proposition 3.2 and Proposition 3.1, we have

ϕ(DD)\displaystyle\phi(D^{\prime}-D) =\displaystyle= FzDD(F)ϕ(F)\displaystyle\sum_{F\in\mathcal{F}}z_{D^{\prime}-D}(F)\phi(F)
=\displaystyle= F(zDD(F)zDD(Fmin))ϕ(F)\displaystyle\sum_{F\in{\mathcal{F}}}(z_{D^{\prime}-D}(F)-z_{D^{\prime}-D}(F_{\min}))\phi(F)
=\displaystyle= F{Fmin}(zDD(F)zDD(Fmin))ϕ(F).\displaystyle\sum_{F\in{\mathcal{F}}\setminus\{F_{\min}\}}(z_{D^{\prime}-D}(F)-z_{D^{\prime}-D}(F_{\min}))\phi(F).

Since zDD(F)zDD(Fmin)=zDD(F)zmin0z_{D^{\prime}-D}(F)-z_{D^{\prime}-D}(F_{\min})=z_{D^{\prime}-D}(F)-z_{\min}\geq 0 for any F{Fmin}F\in{\mathcal{F}}\setminus\{F_{\min}\}, so by Lemma 3.3, DD can be transformed from DD^{\prime} by a sequence of {Fmin}\{F_{\min}\}-forbidden flips.

Let 𝒮{\cal S} be an arbitrary sequence of {Fmin}\{F_{\min}\}-forbidden flips that transform DD^{\prime} into DD. For any facial circuit FF\in{\cal F}, consider the number t(F)t(F) of flips in 𝒮{\cal S} taking on FF. By the definition of 𝒮{\cal S}, we have t(Fmin)=0t(F_{\min})=0.

Similar to the proof of the necessity, (12) is also satisfied. Set K=FminK=F_{\min} in (12). Since t(Fmin)=0t(F_{\min})=0 and zDD(F)=zminz_{D^{\prime}-D}(F)=z_{\min} for every F𝒞F\in{\cal C}, we have t(F)=t(Fmin)=0t(F)=t(F_{\min})=0 for every F𝒞F\in{\cal C}. This implies that 𝒮{\cal S} contains no flip taking on a facial circuit in 𝒞{\cal C}. That is, 𝒮{\cal S} is a 𝒞{\cal C}-forbidden sequence.

Finally, by the arbitrariness of the sequence 𝒮{\cal S}, we may assume that 𝒮{\cal S} is shortest. Thus, by (12) and the fact that t(Fmin)=0t(F_{\min})=0, we have

d(D,D)\displaystyle d(D^{\prime},D) =\displaystyle= Ft(F)=F(t(F)t(Fmin))\displaystyle\sum_{F\in{\cal F}}t(F)=\sum_{F\in{\cal F}}(t(F)-t(F_{\min}))
=\displaystyle= F(zDD(F)zmin).\displaystyle\sum_{F\in{\cal F}}(z_{D^{\prime}-D}(F)-z_{\min}).

This completes our proof. ∎

Corollary 3.5.

Let DD and DD^{\prime} be two α\alpha-orientations of a surface graph. Then DD can be transformed from DD^{\prime} by a sequence of flips if and only if DD and DD^{\prime} are homologous.

Proof.

Recall that zDD(Fmin)=zminz_{D^{\prime}-D}(F_{\min})=z_{\min}. So by Theorem 3.4, if DD and DD^{\prime} are homologous then DD can be transformed from DD^{\prime} by a sequence of {Fmin}\{F_{\min}\}-forbidden flips. The converse is also true by Theorem 3.4. ∎

We now give an extension of Lemma 3.3 as follows.

Corollary 3.6.

Let DD and DD^{\prime} be two α\alpha-orientations of a surface graph and 𝒞{\cal C} be a class of facial circuits. If

ϕ(DD)=F𝒞λFϕ(F),\phi(D^{\prime}-D)=\sum_{F\in\mathcal{F}\setminus{\cal C}}\lambda_{F}\phi(F), (14)

then DD can be transformed from DD^{\prime} by a sequence of 𝒞{\cal C}-forbidden flips if and only if λF0\lambda_{F}\geq 0 for all F𝒞F\in\mathcal{F}\setminus{\cal C}.

Proof.

We write (14) as

ϕ(DD)=FλFϕ(F),\phi(D^{\prime}-D)=\sum_{F\in\mathcal{F}}\lambda_{F}\phi(F), (15)

where λF=0\lambda_{F}=0 if F𝒞F\in{\cal C}. By Proposition 3.2, we have

ϕ(DD)=FzDD(F)ϕ(F).\phi(D^{\prime}-D)=\sum_{F\in\mathcal{F}}z_{D^{\prime}-D}(F)\phi(F).

So by Proposition 3.1, there exists an integer kk such that zDD(F)=λF+kz_{D^{\prime}-D}(F)=\lambda_{F}+k for all FF\in\mathcal{F}. Further, the condition λF0\lambda_{F}\geq 0 for all F𝒞F\in\mathcal{F}\setminus{\cal C} implies that the minimum value of λF\lambda_{F} in (15) is 0 and attained by every F𝒞F\in{\cal C}. Hence, the minimum value zminz_{\min} of zDD(F)z_{D^{\prime}-D}(F) is attained by every F𝒞F\in{\cal C}. The corollary follows directly from Theorem 3.4. ∎

4 Flip graph of α\alpha-orientations

For a class 𝒞{\cal C} of forbidden facial circuits of a surface graph GG, we define the 𝒞{\cal C}-forbidden flip graph of GG as a directed graph, denoted by 𝐃(𝒞){\bf D}({\cal C}), whose vertex set is the class of all α\alpha-orientations of GG, and two orientations DD and DD^{\prime} are joined by a directed edge from DD^{\prime} to DD provided DD can be transformed from DD^{\prime} by a 𝒞{\cal C}-forbidden flip, or equivalently, DDD^{\prime}-D is a ccw facial circuit and DD𝒞D^{\prime}-D\notin{\cal C}. Since the homology is an equivalence relation, the α\alpha-orientations of GG can be partitioned into several equivalence classes, say Ω1,Ω2,,Ωq\Omega_{1},\Omega_{2},\dotsc,\Omega_{q}. For k{1,2,,q}k\in\{1,2,\dotsc,q\}, let 𝐃k(𝒞){\bf D}_{k}({\cal C}) be the subgraph of 𝐃(𝒞){\bf D}({\cal C}) induced by Ωk\Omega_{k}. If 𝒞{\cal C} consists of exactly one facial circuit, then by Theorem 1.1, 𝐃k(𝒞){\bf D}_{k}({\cal C}) is the cover graph of a distributive lattice.

In this section we consider the general case, in which the forbidden class 𝒞{\cal C} consists of any number of facial circuits (in particular, 𝒞{\cal C} may be empty). To this end, we apply the technique of U-coloring and L-coloring, which was introduced by Felsner and Knauer to characterize the cover graph of distributive lattices.

For a directed graph D=(V,E)D=(V,E) and two vertices u,vVu,v\in V, we use (u,v)(u,v) to denote the directed edge joining uu and vv with direction from uu to vv. An edge coloring cc of DD is called a U-coloring if for every u,v,wVu,v,w\in V with uwu\not=w and (v,u),(v,w)E(v,u),(v,w)\in E, the following two rules hold [3]:
(U1). c(v,u)c(v,w)c(v,u)\not=c(v,w);
(U2). There is a vertex zVz\in V and edges (u,z),(w,z)(u,z),(w,z) such that c(v,u)=c(w,z)c(v,u)=c(w,z) and c(v,w)=c(u,z)c(v,w)=c(u,z).

An L-coloring is defined as the dual, in the sense of edge direction reversal, to a U-coloring, that is, every directed edge (i,j)(i,j) in the definition of U-coloring is replaced by (j,i)(j,i).

Proposition 4.1.

[3] If a finite connected acyclic digraph DD admits a U- and an L-coloring then DD is the cover graph of a distributive lattice, and has a unique sink and a unique source.

We now turn to our flip graph. Let ={F1,F2,,F||}{\cal F}=\{F_{1},F_{2},\dotsc,F_{|{\cal F}|}\} be the class of all facial circuits of GG. An edge of 𝐃(){\bf D}(\emptyset) is called of type FiF_{i} if it corresponds to a flip taking on FiF_{i}. Let 𝐄i{\bf E}_{i} be the set of the edges of type FiF_{i} and let 𝐜:E(𝐃()){1,2,,||}{\bf c}:E({\bf D}(\emptyset))\rightarrow\{1,2,\dotsc,|{\cal F}|\} be the edge coloring of 𝐃(){\bf D}(\emptyset) such that 𝐜(𝐞)=i{\bf c}({\bf e})=i for each 𝐞𝐄i,i=1,2,,||{\bf e}\in{\bf E}_{i},i=1,2,\dotsc,|{\cal F}|.

Proposition 4.2.

𝐜{\bf c} is both a U-coloring and an L-coloring.

Proof.

Let D1,D2,D3V(𝐃())D_{1},D_{2},D_{3}\in V({\bf D}(\emptyset)) with D2D3D_{2}\not=D_{3} and (D1,D2)𝐄i,(D1,D3)𝐄j(D_{1},D_{2})\in{\bf E}_{i},(D_{1},D_{3})\in{\bf E}_{j}. Therefore, D2D_{2} (resp., D3D_{3}) can be transformed from D1D_{1} by a flip taking on the ccw facial circuit FiF_{i} (resp., FjF_{j}). Since D2D3D_{2}\not=D_{3}, we have iji\not=j and, hence, the rule U1 holds. Further, since both FiF_{i} and FjF_{j} are ccw, they cannot share the same edge of GG, that is, FiF_{i} and FjF_{j} are independent. Let D4D_{4} be the orientation of GG transformed from D2D_{2} by taking a flip on FjF_{j}. Then D4D_{4} can also be transformed from D3D_{3} by taking a flip on FiF_{i}. This means that (D3,D4)𝐄i,(D2,D4)𝐄j(D_{3},D_{4})\in{\bf E}_{i},(D_{2},D_{4})\in{\bf E}_{j} and, therefore, the rule U2 holds. The discussion for L-coloring is analogous. ∎

For a set E of edges in 𝐃(){\bf D}(\emptyset), we denote by 𝐃()𝐄{\bf D}(\emptyset)-{\bf E} the directed graph obtained from 𝐃(){\bf D}(\emptyset) by deleting all the edges in E. By the definition of the edge coloring 𝐜{\bf c}, the following proposition is obvious.

Proposition 4.3.

For any I{1,2,,||}I\subseteq\{1,2,\dotsc,|{\cal F}|\}, 𝐜{\bf c} restricted on the subgraph 𝐃()iI𝐄i{\bf D}(\emptyset)-\bigcup_{i\in I}{\bf E}_{i} is still a U-coloring and an L-coloring.

Proposition 4.4.

For any facial circuit class 𝒞={Fi1,Fi2,,Fis}{\cal C}=\{F_{i_{1}},F_{i_{2}},\dotsc,F_{i_{s}}\}, we have

𝐃(𝒞)=𝐃()(𝐄i1𝐄i2𝐄is).{\bf D}({\cal C})={\bf D}(\emptyset)-({\bf E}_{i_{1}}\cup{\bf E}_{i_{2}}\cup\cdots\cup{\bf E}_{i_{s}}). (16)
Proof.

Notice that 𝐃(𝒞){\bf D}({\cal C}) and 𝐃()(𝐄i1𝐄i2𝐄is){\bf D}(\emptyset)-({\bf E}_{i_{1}}\cup{\bf E}_{i_{2}}\cup\cdots\cup{\bf E}_{i_{s}}) have the same vertex set. If an edge 𝐞{\bf e} is in some 𝐄ij{\bf E}_{i_{j}} then 𝐞{\bf e} is of type FijF_{i_{j}} and therefore, 𝐞{\bf e} is not in 𝐃(𝒞){\bf D}({\cal C}) since FijF_{i_{j}} is a forbidden facial circuit in 𝒞{\cal C}. The converse is also true and thus (16) follows. ∎

In contrast to the U- and L-colorings for the cover graph of a lattice, we will see that 𝐃(𝒞){\bf D}({\cal C}) may be cyclic if 𝒞={\cal C}=\emptyset, which gives a non-trivial example of cyclic directed graph that admits a U- and an L-coloring. As an example, let us consider the surface graph GG in Example 1. Recall that the equivalence classes, i.e., the homologous classes, of the (2,2,2,2)(2,2,2,2)-orientations of GG are

Ωi={Di}fori{1,2,,12}andΩ13={D13,D14,,D18}.\Omega_{i}=\{D_{i}\}\ \ {\rm for}\ i\in\{1,2,\dotsc,12\}\ \ {\rm and}\ \ \Omega_{13}=\{D_{13},D_{14},\dotsc,D_{18}\}.

This means that the flip graph 𝐃(){\bf D}(\emptyset) of GG consists of 12 isolated vertices D1,D2,,D12D_{1},D_{2},\dotsc,D_{12} and the subgraph 𝐃13(){\bf D}_{13}(\emptyset) induced by Ω13\Omega_{13}, where 𝐃13(){\bf D}_{13}(\emptyset) is illustrated as in Figure 4 and is cyclic. Further, by Proposition 4.4, we have 𝐃13({F4})=𝐃13()𝐄4{\bf D}_{13}(\{F_{4}\})={\bf D}_{13}(\emptyset)-{\bf E}_{4} and 𝐃13({F1,F4})=𝐃13()(𝐄1𝐄4){\bf D}_{13}(\{F_{1},F_{4}\})={\bf D}_{13}(\emptyset)-({\bf E}_{1}\cup{\bf E}_{4}). Moreover, 𝐃13({F4}){\bf D}_{13}(\{F_{4}\}) has one component and 𝐃13({F1,F4}){\bf D}_{13}(\{F_{1},F_{4}\}) has three components, each of which is the cover graph of a distributive lattice, see Figure 4.

[Uncaptioned image]

Figure 4. For i{1,2,3,4}i\in\{1,2,3,4\}, the edges of type FiF_{i} are represented by ii.

Recall that a directed graph DD is strongly connected if for any two vertices uu and vv, there is a directed path from uu to vv and a directed path from vv to uu. The maximum integer ss for which DHD\setminus H is strongly connected for any set HH of at most s1s-1 edges is called the strong edge-connectivity of DD and is denoted by κ(D)\kappa^{\prime}(D).

For k{1,2,,q}k\in\{1,2,\ldots,q\}, let Ok(G,𝒞)O_{k}(G,{\cal C}) be the class of all the α\alpha-orientations in Ωk\Omega_{k} that has no ccw facial circuit other than that in 𝒞{\cal C} and let O(G,𝒞)=k=1qOk(G,𝒞)O(G,{\cal C})=\bigcup_{k=1}^{q}O_{k}(G,{\cal C}). Similarly, let Ok(G,𝒞)O^{*}_{k}(G,{\cal C}) be the class of all the α\alpha-orientations in Ωk\Omega_{k} that has no cw facial circuit other than that in 𝒞{\cal C}. We can see that, for any DOk(G,𝒞)D\in O_{k}(G,{\cal C}), the orientation obtained from DD by reversing the directions of all edges of DD is in Ok(G,𝒞)O^{*}_{k}(G,{\cal C}), and vise versa. This means that |Ok(G,𝒞)|=|Ok(G,𝒞)||O_{k}(G,{\cal C})|=|O^{*}_{k}(G,{\cal C})|. For instance, in the example above we have O13(G,{F1,F4})={D13,D15,D18}O_{13}(G,\{F_{1},F_{4}\})=\{D_{13},D_{15},D_{18}\} and O13(G,{F1,F4})={D14,D15,D18}O^{*}_{13}(G,\{F_{1},F_{4}\})=\{D_{14},D_{15},D_{18}\}, see Figure 2. Further, we can see that D13,D15,D18D_{13},D_{15},D_{18} and D14,D15,D18D_{14},D_{15},D_{18} are exactly the three sinks and three sources of 𝐃13({F1,F4}){\bf D}_{13}(\{F_{1},F_{4}\}), respectively, see Figure 4.

Theorem 4.5.

Let k{1,2,,q}k\in\{1,2,\dotsc,q\}.
1). 𝐃k(){\bf D}_{k}(\emptyset) is strongly connected and κ(𝐃k())=1\kappa^{\prime}({\bf D}_{k}(\emptyset))=1;
2). For any 𝒞{\cal C}\subseteq{\cal F} with 𝒞{\cal C}\not=\emptyset, 𝐃k(𝒞){\bf D}_{k}({\cal C}) has exactly |Ok(G,𝒞)|=|Ok(G,𝒞)||O_{k}(G,{\cal C})|=|O^{*}_{k}(G,{\cal C})| components, each of which is the cover graph of a distributive lattice with the unique sink in Ok(G,𝒞)O_{k}(G,{\cal C}) and a unique source in Ok(G,𝒞)O^{*}_{k}(G,{\cal C}).

Proof.

1). By Corollary 3.5, 𝐃k(){\bf D}_{k}(\emptyset) is strongly connected and therefore, κ(𝐃k())1\kappa^{\prime}({\bf D}_{k}(\emptyset))\geq 1. Let FF be an arbitrary facial circuit FF and let (F){\cal L}(F) be the lattice on the homology class Ωk\Omega_{k} with forbidden circuit FF, where two orientations DD and DD^{\prime} have the relation DDD^{\prime}\leq D in (F){\cal L}(F) if DD can be transformed from DD^{\prime} by a sequence of 𝒞{\cal C}-forbidden flips. Then by Theorem 1.1, 𝐃k(F){\bf D}_{k}(F) is the cover graph of (F){\cal L}(F). Recall that (F){\cal L}(F) has a unique maximal element. Moreover, this maximal element clearly corresponds to an orientation DmaxD_{\max} of GG that contains no ccw facial circuit other than FF [6]. On the other hand, since κ(𝐃k())1\kappa^{\prime}({\bf D}_{k}(\emptyset))\geq 1, DmaxD_{\max} has at least one ccw facial circuit. Thus, FF itself must be ccw in DmaxD_{\max}. This means that FF is the unique ccw facial circuit in DmaxD_{\max} and therefore, κ(𝐃k())=1\kappa^{\prime}({\bf D}_{k}(\emptyset))=1.

2). We first prove that 𝐃k(𝒞){\bf D}_{k}({\cal C}) is acyclic. Suppose to the contrary that 𝐃k(𝒞){\bf D}_{k}({\cal C}) has a directed cycle 𝐂=D1D2DtD1{\bf C}=D_{1}D_{2}\cdots D_{t}D_{1}. Without loss of generality, assume that the facial circuits sequence of the flips corresponding to 𝐂{\bf C} is 𝒮:F1,F2,,Ft{\cal S}:F_{1},F_{2},\dotsc,F_{t}. Since 𝐂{\bf C} is a directed cycle, the direction of every edge in D1D_{1} is not changed after taking the above flips. This means that, for every edge ee in D1D_{1}, the two facial circuits that share ee appear the same times in the sequence 𝒮{\cal S}. Therefore, every facial circuit in D1D_{1} must appear the same times in 𝒮{\cal S} since the dual graph of GG is connected. This is a contradiction because the facial circuits in 𝒞{\cal C} are forbidden to any flip.

Finally, by Proposition 4.2, Proposition 4.3 and Proposition 4.4, 𝐃k(𝒞){\bf D}_{k}({\cal C}) admits a U- and an L-coloring. So by Proposition 4.1, each component of 𝐃k(𝒞){\bf D}_{k}({\cal C}) is the cover graph of a distributive lattice (in particular, a single element) with a unique sink and a unique source. Moreover, if an orientation DD of GG is a sink of 𝐃k(𝒞){\bf D}_{k}({\cal C}), then DD has no ccw facial circuit other than that in 𝒞{\cal C}, meaning that DOk(G,𝒞)D\in O_{k}(G,{\cal C}). The discussion for DD being the source is analogous. This completes our proof. ∎

Acknowledgments

This work is supported by National Natural Science Foundation of China under Grant No. 11971406.

References

  • [1] Y. Disser, J. Matuschke, Degree-constrained orientations of embedded graphs, J. Comb. Optim., 31:758-773, 2016.
  • [2] S. Felsner, Lattice structures from planar graphs, Electron. J. Combin., 11:#R15, 2004.
  • [3] S. Felsner, K. B. Knauer, ULD-Lattices and Δ\Delta-Bonds, Combinatorics, Probability &\& Computing, 18 (5):707-724, 2012.
  • [4] H. de Fraysseix, P. Ossona de Mendez, P. Rosenstiehl, Bipolar orientations revisited, Discrete Appl. Math., 56:157-179, 1995.
  • [5] E. Fusy, Transversal structures on triangulations: A combinatorial study and straight-line drawings, Discrete Math., 309:1870-1894, 2009.
  • [6] D. Gonçalves, K. B. Knauer, B. Lévêque, On the structure of Schnyder woods on orientable surfaces, arXiv: 1501.05475v2, 2016.
  • [7] Joan P. Hutchinson, On short noncontractible cycles in embedded graphs, SIAM J. Discrete Math., 1(2):185-192, 1988.
  • [8] R. Kenyon, J. Propp, D. B. Wilson, Trees and matchings, Electron. J. Combin., 7(1):#R25, 2000.
  • [9] K. B. Knauer, Partial orders on orientations via cycle flips, Ph.D thesis, Technische Universität Berlin, Berlin, 2007.
  • [10] P. C. B. Lam, H.P. Zhang, A distributive lattice on the set of perfect matchings of a plane bipartite graph, Order, 20:13-29, 2003.
  • [11] A. Nakamoto, K. Ota, T. Tanuma, Three-cycle reversions in oriented planar triangulations, Yokohama Math. J., 44:123-139, 1997.
  • [12] J. Propp, Lattice structure for orientations of graphs, arXiv:math/ 0209005v2, 2020.
  • [13] F. J. Zhang, X.F. Guo, R.S. Chen, ZZ-transformation graphs of perfect matchings of hexagonal systems, Discrete Math., 72:405-415, 1988.
  • [14] H. P. Zhang, F.J. Zhang, Plane elementary bipartite graphs, Discrete Appl. Math., 105:291-311, 2000.