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

Genus of Embedded Graphs in Orientable Closed Surfaces

Lorena Armas-Sanabria CONAHCYT-Universidad Autónoma del Estado de Morelos, Av. Universidad 1001, Cuernavaca 62209, Mor., México. [email protected]  and  Víctor Núñez Centro de Investigación en Matemáticas, A.P. 420, Guanajuato 36000, Gto., México. [email protected]
Abstract.

We give an algorithm to calculate the minimal and maximal genus of the orientable closed surface where a graph GG can be embedded. For this, we construct some special branched coverings of the 2-sphere. We apply this algorithm to calculate the orientable genus and maximal genus of some Snarks graphs.

Key words and phrases:
Embedded graphs, genus of a graph, orientable closed surfaces, branch covers of the sphere.
2020 Mathematics Subject Classification:
05C10, 57M15

1. Introduction

The orientable genus of a graph GG is defined as the minimal genus of an orientable closed surface where the graph can be embedded. The maximal orientable genus of GG is the maximal genus of an orientable closed surface where the graph can be cellularly embedded, and remember that an embedding is cellular if the complementary regions of the graph in the surface consist of open disks. The set of numbers bounded by the orientable genus and the maximal orientable genus is called the genus range of GG.

Given a graph G(V,E)G(V,E) where VV is the set of vertices and EE is the set of edges of GG, there is a way to calculate the genus of GG, by using what is called a rotation system [15]. A rotation system consists in an assignment of a numeration to the edges and vertices, in such a way that each vertex has all its incident edges numerated, and then a cyclic permutation is assigned by considering the incident edges to each vertex viv_{i}. It is not difficult to see that this cyclic permutation defines an embedding of GG on an orientable closed surface. This fact was first observed by Edmonds [5]. There are many possibilities of assigning a cyclic permutation to each vertex, these possibilities define all the possible embeddings of the graph in different surfaces. The first algorithm to calculate the genus of a graph was given by Youngs [18], which consists in producing all rotation systems, and for each such system determine the cycles who are to bound a disk in the surface. Then knowing the number of vertices, edges and cycles that bound disks, the genus of the embedding can be calculated by means of the Euler characteristic. This algorithm determines all the genus range of GG, but it requires many steps, for if deg(vi)deg(v_{i}) denotes the degree of vertex viv_{i}, note that there are vV(G)(deg(vi)1)!\prod_{v\in V(G)}(deg(v_{i})-1)! rotation systemss. Also, by a result of Duke [4], it is known that if kk is a number in hte renge genus of GG, then there exists a celular embedding of GG in a closed orientable surface of genus kk.

There are many other methods to determine the genus of a graph, see the survey paper of Stahl [16].

The problem of calculating the genus of a graph is an NPNP-complete problem [17]. There are some bounds for these genera [15], which usually, are not very sharp. A way to show that a graph cannot be embedded in some surface, is determining prohibited minors of graphs for a given orientable, closed surface FF; but it is also a difficult problem to find complete sets of prohibited minors. Given an arbitrary orientable closed surface FF, a complete set of prohibited minors is not known, except for the case FF is a 2-sphere, which is the Kuratowski Theorem, which states that a given graph is planar if and only if it does not have a subgraph homeomorphic to K5K_{5}, or to K3,3K_{3,3}. Also there are some known prohibited minors for the case of the torus, and in the non-orientable case, a complete set of prohibited minors for the projective plane is known [1].

There are several algorithms for embedding graphs of arbitrary genus given by Filotti, Miller and Reif [6], and by Djidjev and Reif [3], but in [12], it is pointed out that they are incorrect and that there is not way to fix them without creating algorithms which take exponential time. Mohar, in [10], [11], proved that if SS is a fixed surface then there is a linear time algorithm to know if an arbitrary graph GG embeds in SS (and finds an embedding) or, finds a subgraph of GG which is a subdivision of some graph in the set of forbidden graphs of SS (see also [13]). However, in [12], it is claimed that this algorithm is very complex and very difficult to implement.

In [7], it is constructed a quadratic-time algorithm for calculating the genus distribution of the graphs in a family 𝔽{\mathbb{F}}, consisting of graphs of fixed treewidth and bounded degree, which complements an algorithm of Kawarabayashi, Mohar and Reed [8] for calculating the minimum genus of a graph of bounded treewidth in linear time.

In this paper we develop an algorithm to calculate the genus and the maximal genus (in fact, the genus range) of a graph, which is equivalent to the construction of the rotation systems of the graph, but in a nicer setting. Our main technique is the construction of certain branched coverings of the 2-sphere. This gives short and clear proofs of the correctness of the algorithm, and also explicit embeddings.

The paper is organized as follows. In Section 2 we recall the elements of branched covering theory. In Section 3 we describe the algorithm to calculate the orientable genera of a graph GG, and we show that the algorithm works. In Section 4 we work an example, and finally in Section 5 there is a table with the orientable genus and maximal orientable genus for some Snarks.

2. Branched covers of the 2-sphere

For a nice account of branched covering theory, see [2]. A finite-to-one, open and proper map φ:MN\varphi:M\rightarrow N between two nn-manifolds is called a branched covering. The usual way to check that an open map φ:MN\varphi:M\rightarrow N is a branched covering, is to find a codimension 2 properly embedded submanifold kNk\subset N such that φ:Mφ1(k)Nk\varphi:M-\varphi^{-1}(k)\rightarrow N-k is a dd-fold covering space. One says that φ\varphi is branched along kk, and that kk is the branching of φ\varphi. We write φ:M(N,k)\varphi:M\rightarrow(N,k) for a branched covering φ:MN\varphi:M\rightarrow N with branching kk.

For a given dd-fold branched covering φ:M(N,k)\varphi:M\rightarrow(N,k), there is an associated representation (that is, a homomorphism) of the fundamental group π1(Nk)\pi_{1}(N-k) into Σd\Sigma_{d}, the symmetric group on dd symbols, ωφ:π1(Nk)Σd\omega_{\varphi}:\pi_{1}(N-k)\rightarrow\Sigma_{d}, as follows: Fix a base point x0Nkx_{0}\in N-k, and number φ1(x0)={x1,,xd}\varphi^{-1}(x_{0})=\{x_{1},\dots,x_{d}\}. For [α]π1(Nk,x0)[\alpha]\in\pi_{1}(N-k,x_{0}), consider the lifting of α\alpha at the point xix_{i} which ends, say, in the point xjx_{j}. Then define ωφ([α])(i)=j\omega_{\varphi}([\alpha])(i)=j.

For a given representation ω:π1(Nk)Σd\omega:\pi_{1}(N-k)\rightarrow\Sigma_{d}, one has an associated dd-fold branched covering φω:M(N,k)\varphi_{\omega}:M\rightarrow(N,k): it is the completion of the covering space M0NkM_{0}\rightarrow N-k corresponding to the subgroup ω1(Σd1)\omega^{-1}(\Sigma_{d-1}).

Equivalence classes of dd-fold connected branched covers M(N,k)M\rightarrow(N,k) are in one-to-one correspondence with conjugacy classes of transitive homomorphisms π1(Nk)Σd\pi_{1}(N-k)\rightarrow\Sigma_{d}.

We are interested in branched coverings of the 2-sphere S2S^{2} branched along three points x1,x2,x3S2x_{1},x_{2},x_{3}\in S^{2}. These coverings are related to a cell decomposition of the 2-sphere consisting of one vertex vv, two edges aa and bb, and three disks D1,D2,D3D_{1},D_{2},D_{3} oriented in such a way that D1=a,D2=b,D3=(ab)1\partial D_{1}=a,\partial D_{2}=b,\partial D_{3}=(a*b)^{-1}. We regard xix_{i} as the center of DiD_{i} (see Figure 1).

Refer to caption
Figure 1. A cellular decomposition of the 2-sphere.

The fundamental group of the punctured 2-sphere is free of rank two and admits a presentation π1(S2{x1,x2,x3})x¯1,x¯2,x¯3:x¯1x¯2x¯3=1\pi_{1}(S^{2}-\{x_{1},x_{2},x_{3}\})\cong\langle\bar{x}_{1},\bar{x}_{2},\bar{x}_{3}:\bar{x}_{1}\bar{x}_{2}\bar{x}_{3}=1\rangle, where the symbol x¯i\bar{x}_{i} is the class of a closed curve on S2S^{2} starting at the vertex vv and encircling the point xix_{i}. Alternatively, using the cell decomposition of S2S^{2} above, we obtain a presentation π1(S2{x1,x2,x3})a,b,c:abc=1\pi_{1}(S^{2}-\{x_{1},x_{2},x_{3}\})\cong\langle a,b,c:abc=1\rangle where ax¯1a\simeq\bar{x}_{1}, bx¯2b\simeq\bar{x}_{2}, and c=(ab)1x¯3c=(a*b)^{-1}\simeq\bar{x}_{3}. Therefore an dd-fold branched covering of S2S^{2} branched along {x1,x2,x3}\{x_{1},x_{2},x_{3}\} is determined by a triple of permutations α,β,γΣd\alpha,\beta,\gamma\in\Sigma_{d}, such that αβγ=1\alpha\beta\gamma=1. The triple (α,β,γ)(\alpha,\beta,\gamma) is called a constellation in [9]. An arbitrary pair of permutations α,βΣd\alpha,\beta\in\Sigma_{d} gives a constellation, (α,β,γ)(\alpha,\beta,\gamma), setting γ=(αβ)1\gamma=(\alpha\beta)^{-1}.

A constellation α,β,γΣd\alpha,\beta,\gamma\in\Sigma_{d} defines a representation ω:π1(S2{x1,x2,x3})Σd\omega:\pi_{1}(S^{2}-\{x_{1},x_{2},x_{3}\})\rightarrow\Sigma_{d} such that ω(a)=α\omega(a)=\alpha, ω(b)=β\omega(b)=\beta, and ω(c)=γ\omega(c)=\gamma. The associated branched covering φω:FS2\varphi_{\omega}:F\rightarrow S^{2} can be constructed in two steps. First, construct the covering space of aba\vee b induced by the restriction ω:π1(ab)Σd\omega:\pi_{1}(a\vee b)\rightarrow\Sigma_{d}, and, secondly, construct the branched coverings of the disks DiD_{i} branched along xix_{i} associated to the restrictions ω:π1(Di{xi})Σd\omega:\pi_{1}(D_{i}-\{x_{i}\})\rightarrow\Sigma_{d}. These covers have as many components as the number of cycles in a disjoint cycle decomposition of ω((Di))\omega(\partial(D_{i})). Finally assemble together these coverings into FF with liftings of the glueing maps Diab\partial D_{i}\rightarrow a\vee b.

The dd-fold branched cover ϕ:FS2\phi:F\rightarrow S^{2} gets a cell decomposition for FF induced by the cell decomposition of S2S^{2}, with vertices φ1(v)\varphi^{-1}(v), edges φ1(ab)\varphi^{-1}(a\cup b), and 2-cells φ1(D1D2D3)\varphi^{-1}(D_{1}\cup D_{2}\cup D_{3}).

3. Orientable genera of graphs

In this section we describe an algorithm to calculate the orientable genera of a graph GG. For a set YY, we write Σ(Y)\Sigma(Y) for the symmetric group in the symbols in YY. If #(Y)=k\#(Y)=k, we write Σk=Σ(Y)\Sigma_{k}=\Sigma(Y). If γΣk\gamma\in\Sigma_{k}, |γ||\gamma| is the number of cycles in a decomposition of γ\gamma into disjoint cycles in Σk\Sigma_{k} (including trivial cycles for the fixed points of γ\gamma).

3.1. The algorithm


We start with a finite graph GG with vertices v1,,vnv_{1},\cdots,v_{n}, and edges f1,,fef_{1},\cdots,f_{e}. Without loss of generality, we assume GG simple and with all vertices of degree at least 3.

1 ) Add new fake vertices w1,,wew_{1},\cdots,w_{e} where each vertex wiw_{i} is the middle point of fif_{i}. We obtain two new edges for each fif_{i}, which are called darts in [9] (See Figure 2).

Refer to caption
Figure 2. K4K_{4} with darts.

2 ) Label each dart with a number in {1,2,,2e}\{1,2,\cdots,2e\}, and define the permutation α=(d1,d1)(d2,d2)(de,de)S2e\alpha=(d_{1},{d^{\prime}}_{1})(d_{2},{d^{\prime}}_{2})\cdots(d_{e},{d^{\prime}}_{e})\in S_{2e}, where di,did_{i},{d^{\prime}}_{i} are the numbers of the darts of the original edge fif_{i}.

3 ) For each vertex viv_{i}, consider cic_{i}, the conjugacy class of the tit_{i}-cycle (si,1,si,2,,si,ti)(s_{i,1},s_{i,2},\dots,s_{i,t_{i}}) in Σ({si,1,si,2,,si,ti}){\Sigma(\{s_{i,1},s_{i,2},\dots,s_{i,t_{i}}\})}, where si,1,si,2,,si,tis_{i,1},s_{i,2},\dots,s_{i,t_{i}} are the numbers of the darts incident in viv_{i}.

4 ) For each choice β1c1,β2c2,,βncn\beta_{1}\in c_{1},\beta_{2}\in c_{2},\dots,\beta_{n}\in c_{n}, define the permutation β=β1β2βn\beta=\beta_{1}\beta_{2}\cdots\beta_{n}.

5 ) Collect the numbers (2(|β||α|+|(αβ)1|))/2(2-(|\beta|-|\alpha|+|(\alpha\beta)^{-1}|))/2 in a set XX, for all possible α,β\alpha,\beta.

Note that each element of a set cic_{i} is a permutation system around the vertex viv_{i}, and a choice of permutation β\beta is just a choice of a rotation system for the graph.

Let g(G)g(G) and gM(G)g_{M}(G) be the orientable genus and the maximal orientable genus of GG, respectively.


Proposition 3.1.

Let G(V,E)G(V,E) be a graph such that deg(vi)3deg(v_{i})\geq 3 for all ii. Let X={(2(|β||α|+|(αβ)1|))/2|X=\{(2-(|\beta|-|\alpha|+|(\alpha\beta)^{-1}|))/2\hskip 5.0pt|\hskip 5.0pt for all possible α,β}\alpha,\beta\}. Then g(G)=min(X)g(G)=\min(X) and gM(G)=max(X)g_{M}(G)=\max(X).

3.2. A branched covering

We verify that the described algorithm works. For that we show that the genera of all orientable surfaces FF such that GG embeds in FF, are collected in the set XX above.

Proposition 3.2.

Let G(V,E)G(V,E) be a graph such that deg(vi)3deg(v_{i})\geq 3 for all ii. Then the genus range of GG is given in the set XX.

Proof. We consider the graph GG with its edges subdivided in darts as before. Assume the graph GG is cellularly embedded in an oriented surface FF, and let UU be a regular neighborhood of GG in FF.

Inside UU, substitute each vertex viv_{i} with a fat vertex which is a polygon of tit_{i} vertices where tit_{i} is the valency of the vertex viv_{i}. We visualize this fat vertex as centered in viv_{i}, and intersecting the graph GG in tit_{i} points of the darts incident in viv_{i}. As an example of this construction, we can see Figure 3.

The vertices of this tit_{i}-gon are labeled with the numbers of the incident darts in viv_{i}.

Also, inside UU, substitute each edge fif_{i} of GG with two parallel edges whose ends are the intersection points of fif_{i} with the polygons corresponding to the ends of fif_{i}.

We obtain a graph GG^{\prime} inside UU, whose vertices are the vertices of all of the tit_{i}-gons, and whose edges are the edges of the tit_{i}-gons and the added parallel edges, one pair for each original edge of GG.

This graph GG^{\prime} defines a pair of permutations, α=(d1,d1)(d2,d2)(de,de)\alpha=(d_{1},d_{1}^{\prime})(d_{2},d_{2}^{\prime})\cdots(d_{e},d_{e}^{\prime}) where di,did_{i},d_{i}^{\prime} are the ends of the parallel edges corresponding to the original edge fif_{i}, and β=(s1,1,s1,2,,s1,t1)(s2,1,s2,2,,s2,t2)(sn,1,sn,2,,sn,tn)\beta=(s_{1,1},s_{1,2},\dots,s_{1,t_{1}})(s_{2,1},s_{2,2},\dots,s_{2,t_{2}})\cdots(s_{n,1},s_{n,2},\dots,s_{n,t_{n}}) where si,1,si,2,,si,tis_{i,1},s_{i,2},\dots,s_{i,t_{i}} are the numbers of the vertices of the tit_{i}-gon corresponding to the original vertex viv_{i} and are cyclically ordered by the local orientation of FF at viv_{i}.

This determines a (2e)(2e)-fold covering space ψ:G~ab\psi:\tilde{G}\rightarrow a\vee b corresponding to the representation π1(ab)Σ2e\pi_{1}(a\vee b)\rightarrow\Sigma_{2e} such that aαa\mapsto\alpha, and bβb\mapsto\beta. Notice that in this covering, the preimage of aa has two edges for each transposition (di,di)(d_{i},d_{i}^{\prime}) of α\alpha, and the preimage of bb has a tit_{i}-gon for each cycle (si,1,si,2,,si,ti)(s_{i,1},s_{i,2},\cdots,s_{i,t_{i}}) of β\beta. That is, G~=G\tilde{G}=G^{\prime}.

We can extend ψ\psi into a (2e)(2e)-fold branched covering φ:F(S2,{x1,x2,x3})\varphi:F\rightarrow(S^{2},\{x_{1},x_{2},x_{3}\}) corresponding to the representation π1(S2{x1,x2,x3})a,b,c:abc=1Σ2e\pi_{1}(S^{2}-\{x_{1},x_{2},x_{3}\})\cong\langle a,b,c:abc=1\rangle\rightarrow\Sigma_{2e} such that aαa\mapsto\alpha, bβb\mapsto\beta, and cγ=(αβ)1c\mapsto\gamma=(\alpha\beta)^{-1}.

For a permutation ηΣn\eta\in\Sigma_{n}, we write o(η)o(\eta) for the order of η\eta in the group Σn\Sigma_{n}. Write α=α1αe\alpha=\alpha_{1}\cdots\alpha_{e}, β=β1βn\beta=\beta_{1}\cdots\beta_{n}, and γ=γ1γm\gamma=\gamma_{1}\cdots\gamma_{m} for disjoint cycle decompositions of α,β\alpha,\beta and γ\gamma in Σ2e\Sigma_{2e}. Then αi=(di,di)\alpha_{i}=(d_{i},d^{\prime}_{i}), and βi=(si,1,si,2,,si,ti)\beta_{i}=(s_{i,1},s_{i,2},\cdots,s_{i,t_{i}}). Let χ(F)\chi(F) be the Euler characteristic of FF. Since FF is obtained as a branched covering of S2S^{2}, by the Riemann-Hurwitz formula we have:

χ(F)=2eχ(S2)i=1e(o(αi)1)j=1n(o(βj)1)k=1m(o(γk)1))=4e((i=1e2)e)((j=1ntj)n)((k=1mo(γk))|γ|)=4e(2ee)(2en)(2e|γ|)=ne+|γ|=|β||α|+|(αβ)1|.\begin{array}[]{rl}\displaystyle\chi(F)&\displaystyle=2e\chi(S^{2})-\sum_{i=1}^{e}(o(\alpha_{i})-1)-\sum_{j=1}^{n}(o(\beta_{j})-1)-\sum_{k=1}^{m}(o(\gamma_{k})-1))\\ &\displaystyle=4e-((\sum_{i=1}^{e}2)-e)-((\sum_{j=1}^{n}t_{j})-n)-((\sum_{k=1}^{m}o(\gamma_{k}))-|\gamma|)\\ &=4e-(2e-e)-(2e-n)-(2e-|\gamma|)\\ &=n-e+|\gamma|\\ &=|\beta|-|\alpha|+|(\alpha\beta)^{-1}|.\\ \end{array}

Since χ(F)=22g(F)\chi(F)=2-2g(F), we see that

g(F)=2(|β||α|+|(αβ)1|)2,g(F)=\frac{2-(|\beta|-|\alpha|+|(\alpha\beta)^{-1}|)}{2},

as needed.

Notice that, if we change the orientation of FF, the permutations α\alpha, β\beta and γ\gamma constructed, are just the inverses of the permutations of the text, and, therefore, have the same cycle structure in Σ2e\Sigma_{2e}, giving the same formula for g(F)g(F).

Also notice that, if gg is an arc in the 2-sphere connecting x1x_{1} and x2x_{2} and intersecting aba\cup b exactly in the vertex vv (see Figures 1, 3), then φ1(g)=G\varphi^{-1}(g)=G.

On the other hand, if we are given a pair of permutations α\alpha and β\beta, we can construct a surface FF containing GG, such that the corresponding associated permutations are precisely α\alpha and β\beta. To see this, consider an oriented polygon tit_{i} for each vertex viv_{i}, which carries the permutation β\beta restricted to viv_{i}. For an edge fkf_{k} connecting vertices viv_{i} and vjv_{j}, attach a rectangle connecting the polygons tit_{i} and tjt_{j}, which preserves the orientation of the polygons. We can assume that the union of the disks and the band contains the edge fkf_{k}. Doing this for all edges, produces a surface with boundary, and then by attaching disks to each of its boundary components we get a closed surface FF in which GG embeds cellularly. ∎

Now, just see that Proposition 3.1 is a corollary of Proposition 3.2.

4. Example

In this Section, we apply the algorithm to K4K_{4}, the complete graph of 44 vertices as in Figure 2.

From Figure 2, we obtain the permutation α=(1,9)(2,12)(3,4)(5,11)(6,7)(8,10)\alpha=(1,9)(2,12)(3,4)(5,11)(6,7)(8,10). We choose, as in the figure, β=(1,2,3)(4,5,6)(7,8,9)(10,11,12)\beta=(1,2,3)(4,5,6)(7,8,9)(10,11,12), and then γ=(αβ)1=(1,4,7)(2,9,10)(3,12,5)(6,11,8)\gamma=(\alpha\beta)^{-1}=(1,4,7)(2,9,10)(3,12,5)(6,11,8). Here e=6e=6, tj=3t_{j}=3 for j=1,2,3,4j=1,2,3,4, and lk=3l_{k}=3 for k=1,2,3,4k=1,2,3,4.

Using the Riemann-Hurwitz formula we obtain:

g(X~)=2(|β||α|+|(αβ)1|)/2=2(46+4)/2=0g(\tilde{X})=2-(|\beta|-|\alpha|+|(\alpha\beta)^{-1}|)/2=2-(4-6+4)/2=0

So, X~\tilde{X} is the 22-sphere. But we did know that K4K_{4} is planar.

The graph GG^{\prime} looks like the graph in the left side of Figure 3, and in that figure one can see the corresponding branched cover of the 2-sphere.

Now gM(K4)=1g_{M}(K_{4})=1, and this genus can be achieved with the representation α=(1,9)(2,12)(3,4)(5,11)(6,7)(8,10)\alpha=(1,9)(2,12)(3,4)(5,11)(6,7)(8,10) as before, but β=(1,2,3)(4,5,6)(7,8,9)(10,12,11)\beta=(1,2,3)(4,5,6)(7,8,9)(10,12,11). Then (αβ)1=(4,7,1)(10,5,3,12,8,6,11,2,9)(\alpha\beta)^{-1}=(4,7,1)(10,5,3,12,8,6,11,2,9), and using the genus formula again we obtain:

g(X~)=(2(|β||α|+|(αβ)1|))/2=(2(46+2))/2=1g(\tilde{X})=(2-(|\beta|-|\alpha|+|(\alpha\beta)^{-1}|))/2=(2-(4-6+2))/2=1
Refer to caption
Figure 3. The 2-sphere as a branched cover of the 2-sphere.

5. Orientable genus and maximal orientable genus of some Snarks

In this section, we list the orientable genus and maximal orientable genus of some Snarks. Recall that a Snark is a 3-regular graph, which is 4-edge connected and it is not 3-edge-colourable. Many examples of Snarks are known, this is an interesting but mysterious class of graphs. We use the list of snarks and the notation given in [14]. A program for 3-regular graphs implementing the algorithm of Section 3, and written in the GAP programming language is available from the authors on request.

GraphGraph Orientable genus Maximal or. genus
Petersen 1 3
Tietze 1 3
Blanusa 1 2 5
Blanusa 2 1 5
j5 2 5
Sn13 2 6
Sn18 3 6
Sn19 2 6
Sn20 2 6
Sn21 2 6
Sn22 2 6
Sn23 2 6
Sn24 2 6
Sn25 2 6
Sn26 2 6
Sn27 2 6
Sn28 2 6
Sn29 2 6
Sn27bis 2 6
Sn28bis 2 6
CelminsSwart1 2 7
LoupequineTypeI 1 7
Type2 2 7
CelminsSwart2 2 7
FlowerJ7 2 7
doubleStar 3 8
Double St 3 8
Type1 1 9
Type2 34 2 9
FlowerJ9 2 9

Acknowledgement. The first author is supported by a fellowship Investigadoras por México from CONAHCYT. Research partially supported by the grant PAPIIT-UNAM IN117423.

References

  • [1] D. Archdeacon, A Kuratowski theorem for the projective plane. J. Graph Theory 5 (1981), 243-246.
  • [2] I. Berstein and A. Edmonds, On the construction of branched coverings of low-dimensional manifolds. Trans. Amer. Math. Soc. 247 (1979), 87-124.
  • [3] H. Djidjev and J. Reif, An efficient algorithm for the genus problem with explicit construction of forbidden subgraphs. Proc. of the 23rd. ACM Symp. theory of Comp., (1991), 337-347.
  • [4] R.A. Duke, The genus, regional number, and Betti number of a graph. Canad. J. Math. 18, 817-822.
  • [5] J.R. Edmonds, A combinatorial representation for polyhedral surfaces, American Mathematical Society Notices 7 (1960), 646.
  • [6] I. S. Filotti, G. L. Miller and J. Reif, On determining the genus of a graph in O(vO(g))O(v^{O(g)}) steps. Proc. of the 11th ACM Symp. Theory of Comp., (1979), 27-37.
  • [7] J. L. Gross, Embeddings of graphs of fixed treewidth and bounded degree. ARS Mathematica Contemporanea 7 (2014), 379-403.
  • [8] K. Kawarabayashi, B. Mohar and B. Reed, A simpler linear time algorithm for embedding graphs into an arbitrary surface and the genus of graphs of bounded treewidth. Proc. 49th Ann. Symp. on Foundations of Computer Science (FOCS’08) IEEE (2008), 771-780.
  • [9] S. K. Lando and A. K. Zvonkin, Graphs on Surfaces and their Applications. Encyclopaedia of Mathematical Sciences Vol. 141, Springer-Verlag, 2004.
  • [10] B. Mohar, Embedding graphs in an arbitrary surface in linear time. Proc. 28th Ann. ACM STOC, Philadelphia, ACM Press, 1996, pp. 392-397.
  • [11] B. Mohar, A linear time algorithm for embedding graphs in an arbitrary surface. SIAM J. Discrete Math 12 (1999), 6-26.
  • [12] W. Myrvold and W. Kocay, Errors in graph embedding algorithms. Journal of Computer and System Sciences 77 (2011), 430-438.
  • [13] B. Mohar and C. Thomassen, Graphs on Surfaces. Johns Hopkins University Press, Baltimore, MD, 2001.
  • [14] R. C. Read, R. J. Wilson, An Atlas of Graphs. Oxford Science Publications, 2005.
  • [15] G. Ringel. Map Color Theorem, Die Grundlehren der mathematischen Wissenschaften 209 Berlin-Heidelberg-New York: Springer-Verlag XII, 191 pp. 1974.
  • [16] S. Stahl, The embedding of s graph - a survey. Journal of Graph Theory 2 (1978), 275–298.
  • [17] C. Thomassen, The graph genus problem is NP-complete. J. Algorithms 10 (1989), 568-576.
  • [18] J.W.T. Youngs, Minimal imbeddings and the genus of a graph. J. Math. Mech. 12 (1963), 303–315