
Let G be a graph that can be embedded in a surface Σ of Euler characteristic c′(Σ). In this paper, we proved that there exists an integer Δ0=4⋅(5+√49−24c′)⋅[2−c′+12(5+√49−24c′)] such that the total chromatic number of G is Δ(G)+1 if Δ(G)≥Δ0.
Citation: Qiming Fang, Li Zhang. Graphs that embedded in any fixed surfaces with sufficiently large maximum degree Δ is total-(Δ+1)-colorable[J]. AIMS Mathematics, 2024, 9(1): 1523-1534. doi: 10.3934/math.2024075
[1] | Fugang Chao, Donghan Zhang . Neighbor sum distinguishing total choice number of IC-planar graphs with restrictive conditions. AIMS Mathematics, 2023, 8(6): 13637-13646. doi: 10.3934/math.2023692 |
[2] | Hongyu Chen, Li Zhang . A smaller upper bound for the list injective chromatic number of planar graphs. AIMS Mathematics, 2025, 10(1): 289-310. doi: 10.3934/math.2025014 |
[3] | Chao Yang, Bing Yao, Zhi-xiang Yin . A new vertex distinguishing total coloring of trees. AIMS Mathematics, 2021, 6(9): 9468-9475. doi: 10.3934/math.2021550 |
[4] | Yinwan Cheng, Chao Yang, Bing Yao, Yaqin Luo . Neighbor full sum distinguishing total coloring of Halin graphs. AIMS Mathematics, 2022, 7(4): 6959-6970. doi: 10.3934/math.2022386 |
[5] | Zepeng Li . A note on the bounds of Roman domination numbers. AIMS Mathematics, 2021, 6(4): 3940-3946. doi: 10.3934/math.2021234 |
[6] | Damchaa Adiyanyam, Enkhbayar Azjargal, Lkhagva Buyantogtokh . Bond incident degree indices of stepwise irregular graphs. AIMS Mathematics, 2022, 7(5): 8685-8700. doi: 10.3934/math.2022485 |
[7] | Xiaoxue Hu, Jiangxu Kong . An improved upper bound for the dynamic list coloring of 1-planar graphs. AIMS Mathematics, 2022, 7(5): 7337-7348. doi: 10.3934/math.2022409 |
[8] | Yuehua Bu, Qi Jia, Hongguo Zhu, Junlei Zhu . Acyclic edge coloring of planar graphs. AIMS Mathematics, 2022, 7(6): 10828-10841. doi: 10.3934/math.2022605 |
[9] | Yufei Huang, Hechao Liu . Bounds of modified Sombor index, spectral radius and energy. AIMS Mathematics, 2021, 6(10): 11263-11274. doi: 10.3934/math.2021653 |
[10] | Baolin Ma, Chao Yang . Distinguishing colorings of graphs and their subgraphs. AIMS Mathematics, 2023, 8(11): 26561-26573. doi: 10.3934/math.20231357 |
Let G be a graph that can be embedded in a surface Σ of Euler characteristic c′(Σ). In this paper, we proved that there exists an integer Δ0=4⋅(5+√49−24c′)⋅[2−c′+12(5+√49−24c′)] such that the total chromatic number of G is Δ(G)+1 if Δ(G)≥Δ0.
We consider undirected, finite and simple graphs here. Definitions and notations not given here may be found in [1,2]. For a vertex v, we use dG(v), NG(v) (or simply, d(v), N(v)) to denote the degree of v and the neighborhood of v; furthermore, ∀x∈V(G)∪E(G), and we use NEG(x) to denote the set that contains every edge that is incident with or adjacent to x. For a graph G and a vertex set S, G[S] is the subgraph induced by S. Surfaces in this paper are compact, connected, two-dimensional manifolds without boundary. All embeddings considered in this paper are two-cell embeddings. Let Sk be the orientable surface with k handles and let Nk be the nonorientable surface with k cross-caps. (Note that k is the genus of the surfaces.)
The Euler characteristic of a surface Σ, denoted c′(Σ), is defined by:
c′(Σ)={2−2kifΣishomeomorphictoSk,2−kifΣishomeomorphictoNk. |
For a plane graph G, we use V(G), E(G), Δ(G), δ(G) and F(G) to denote its vertex set, edge set, maximum degree, minimum degree and face set, respectively. A k+-vertex (k−-vertex) is a vertex v with a degree of at least k (at most k). Let f be a face of G and d(f) means the number of vertices that is incident with f. Similarly, we can get the definitions of k+-face and k−-face. The boundary of a t-face f is denoted by ∂(f)=[v1,⋯,vt], the set in which each vertex is incident with f is denoted by V(f).
A proper l-coloring of a graph G is a mapping c from V(G) to the color set {1,2,...,l} such that no two adjacent vertices are assigned the same color. We say G is l-colorable if G has a proper l-coloring. We use c(v) to denote the color of the vertex v. The chromatic number χ(G) of G is the smallest integer k such that G is k-colorable.
A total k-coloring of a graph G is a mapping ϕ from V(G)∪E(G) to the set of colors {1,2,...,k} such that ϕ(x)≠ϕ(y) for every pair of adjacent or incident elements x,y∈V(G)∪E(G). A graph G is total-k-colorable if it has a total-k-coloring. The total chromatic number χ″(G) of G is the smallest integer k such that G is total-k-colorable.
Let L be a list assignment of G; a total coloring c of G is called a total-L-coloring of G if c(v)∈L(v) for any v∈V(G)∪E(G). A graph G is total-k-choosable if G has a total-L-coloring for any list L with |L(v)|=k for each v∈V(G)∪E(G). The total choice number of G, denoted by χ″l(G), is the least integer k such that G is total-k-choosable.
Conjecture 1. Every graph satisfies χ″(G)≤Δ(G)+2.
Conjecture 1 which is called the total coloring conjecture, was given by Behzad [3] and Vizing [15] independently and has only been confirmed for Δ≤5 (See [9,10,11]).
For planar graphs, the bound χ″(G)≤Δ+2 was first proved in 1987 by Borodin [4] for Δ≥11, then for Δ≥9 [5] and then the restriction on Δ was strengthened to Δ≥8 by Jensen and Toft [8] and to Δ≥7 by Sanders and Zhao [14].
Moreover, the bound χ″(G)=Δ+1 was first proved in 1987 by Borodin [4] for Δ≥16, then for Δ≥14 [5] and then the restriction on Δ was improved to 12 and 11 (Borodin, Kostochka and Woodall [6,7]), 10 (Wang [16]) and nine (Kowalik, Sereni and Škrekovski [12]). Wang et al. (2014) strengthened this result and got the following theorem.
Theorem 1.1. (Wang et al. [17]) Let G be a graph that can be embedded in a surface Σ of Euler characteristic c′(Σ)≥0. If Δ≥9, then χ″(G)≤Δ+1.
There are some open problems about total coloring.
Problem 1. (Sanders and Zhao [14]) Is every planar graph with Δ=6 total-8-colorable?
Problem 2. (Kowalik, Sereni, ˇSkrekovski [12]) What is the smallest Δ0, such that every planar graph with Δ(G)≥Δ0 satisfies χ″(G)=Δ(G)+1?
We notice that there are many results for total coloring of graphs that can be embedded in the surfaces of nonnegative Euler characteristics, but none for the surfaces of negative Euler characteristics. Therefore, we give the following conjecture.
Conjecture 2. Let G be a graph that can be embedded in a surface Σ of arbitrary Euler characteristic c′(Σ), then there exists an integer Δ0 such that χ″(G)=Δ(G)+1 if Δ(G)≥Δ0.
In this paper, we prove the following theorem. For the convenience of the following part, let ϵ=12(5+√49−24c′).
Theorem 1.2. Let G be a graph that can be embedded in a surface Σ of Euler characteristic c′<0, then χ″(G)=Δ(G)+1 if Δ(G)≥8ϵ(2−c′+ϵ).
Therefore, Conjecture 2 is the corollary of Theorems 1.1 and 1.2.
Let Sk be the complete bipartite graph K1,k. It is easy to verify that χ″(Sk)=k+1. Let G be a graph with Δ(G)=k, then Sk is a subgraph of G, which implies that χ″(G)≥χ″(Sk)=k+1=Δ(G)+1. Therefore, we only need to prove χ″(G)≤Δ(G)+1 in the following part.
Heawood proved the following analogue of the four color theorem for general surfaces.
Lemma 2.1. (Heawood [13]) Let G be a graph that can be embedded in a surface Σ of Euler characteristic c′≠2 (Σ is not homeomorphic to the sphere), then χ(G)≤12(7+√49−24c′), and G has a vertex of a degree less or equal to 12(5+√49−24c′).
Lemma 2.1 implies that G contains an ϵ−-vertex if G can be embedded in a surface Σ of Euler characteristic c′ (c′≠2 and ϵ=12(5+√49−24c′)).
In order to prove Theorem 1.2, we need the following lemma.
Lemma 2.2. (The Jordan curve theorem [2]) Any simple closed curve C in the plane partitions the rest of the plane into two disjoint arcwise-connected open sets.
We have the following corollary by Lemma 2.2.
Corollary 1. If C1,C2,...,Ci are i simple closed curves in the plane, then C1,C2,...,Ci partition the rest of the plane into at least i+1 disjoint arcwise-connected open sets.
Lemma 2.3. ([1]) Let Sk be the orientable surface with genus g, and p be a point in Sk, then we can find 2g simple closed curves {C1,C2,...,C2g}, such that p∈⋂2gi=1Ci and Sk∖(⋃2gi=1Ci) is homeomorphic to the plane. (Note that these simple closed curves are not homotopic to each other, which impiles that these simple closed curves belong to different generators of the fundamental group π(Sk,p) of Sk.)
Lemma 2.4. ([1]) Let Nk be the nonorientable surface with genus g, and p be a point in Nk, then we can find g simple closed curves {C1,C2,...,Cg}, such that p∈⋂gi=1Ci and Nk∖(⋃gi=1Ci) is homeomorphic to the plane. (Note that these simple closed curves are not homotopic to each other, which impiles that these simple closed curves belong to different generators of the fundamental group π(Nk,p) of Nk.)
We get the following corollary by Lemmas 2.3 and 2.4 and the definition of the Euler characteristic.
Corollary 2. Let Σ be a surface of Euler characteristic c′, then we can find 2−c′ simple closed curves {C1,C2,...,C2−c′} such that these curves have a common point p∈⋂2−c′i=1Ci and Σ∖(⋃2−c′i=1Ci) is homeomorphic to the plane. (For example, four simple closed curves partition S2 into a disc as shown in Figure 1.)
We get the following corollary by Corollaries 1 and 2.
Corollary 3. Let Σ be a surface of Euler characteristic c′, then (2−c′+i) simple closed curves {C1,C2,...,C2−c′+i} in Σ partition the rest of Σ into at least i+1 disjoint arcwise-connected open sets (i∈N).
Proof. We know that 2−c′ simple closed curves can partition Σ into a plane by Corollary 2 (note that Σ remains arcwise-connected in this process. If we do not partition Σ by the method in Corollary 2, Σ will be divided into more pieces and we will get an easier proof), while the remaining i curves further partition the plane into i+1 disjoint arcwise-connected open sets by Corollary 1.
Let G0 be a counterexample of Theorem 1.2, then Δ(G0)≥8ϵ(2−c′+ϵ) and χ″(G0)>Δ(G0)+1. Let S={G:Δ(G)≤Δ(G0)=Δ,χ″(G)>Δ(G0)+1=Δ+1}; the set isn't empty since G0∈S. Let G be the graph with the fewest edges in S. We assume that G has been embedded in the surface Σ.
We prove Theorem 1.2 by constructing a simple graph G′, which can be embedded in Σ from G such that δ(G′)≥ϵ+1, that contradicts the fact that every simple graph that is embedded in surface Σ has an ϵ−-vertex.
Let [Δ+1]={1,2,...,Δ+1} be a color set such that G isn't total-(Δ+1)-colorable. Let c be a coloring of G; we use c(x) to denote the color of x for any x∈V(G)∪E(G). Let X⊆V(G)∪E(G) and define c(X)={c(x):x∈X}.
Let x be a cut vertex of G and let the components of G−x have vertex sets V1, V2, ..., Vt, then the subgraphs Gi=G[Vi∪{x}] are the x-components of G, where i=1,2,…,t.
If P=v1v2...vx−1vx is a path with V(P)={v1,v2,…,vx−1,vx}, then v1 and vx are called the end vertices of P and vi∈{v2,…,vx−1} is called an inner vertex of P.
The length of P, which is denoted by |P|, is the number of edges in P.
Let v1,vx∈V(G), the distance between v1 and vx, denoted dG(v1,vx), be the length of the shortest path from v1 to vx.
A vertex v is big if d(v)≥4ϵ(2−c′+ϵ)+1; otherwise, v is small. The sets of big and small vertices of G are denoted by B(G) and S(G), respectively.
As shown in Figure 2, let C∗ be a cycle of G with the edge set E(C∗). We can regard C∗ as a simple closed curve of the surface Σ. If C partitions the rest of the surface Σ into two disjoint arcwise-connected open sets, the two open sets into which C∗ partitions Σ are called the interior and the exterior of C∗. We denote them by int(C∗) and ext(C∗), and their closures by Int(C∗) and Ext(C∗), respectively (thus, Int(C∗)∩Ext(C∗)=C∗). Without loss of generality, we assume that int(C∗) is homeomorphic to a disc. A vertex v is an inner vertex of C∗ if v∈int(C∗). A vertex v is an outer vertex of C∗ if v∈ext(C∗), and the sets of inner and outer vertices of C∗ are denoted by V(int(C∗)) and V(ext(C∗)), respectively.
Lemma 3.1. G is connected, and δ(G)≥2.
Proof. Suppose to the contrary that G has a 1-vertex v. By the choice of G, G−v has a total-(Δ+1)-coloring c. Let NG(v)={u}, then uv can receive any color except for c(u) and those colors in c(NEG(u)), so the number of forbidden colors of uv is at most 1+(Δ−1)=Δ. Similarly, the number of forbidden colors of v is at most 1+1=2. Therefore, we can extend c to the whole graph G, a contradiction.
Lemma 3.2. Let {u,v}⊆B(G), then the number of 2-vertices, which are adjacent to both u and v, is at most one.
Proof. Suppose to the contrary that there exists two vertices w1 and w2 such that wi is adjacent to both u and v and d(wi)=2 (i∈{1,2}). By the choice of G, G−{w1,w2} has a total-(Δ+1)-coloring c, then the edge uwi can receive any color except for c(u) and those colors in c(NEG(u)). The edge vwi can receive any color except for c(v) and those colors in c(NEG(v)), then the number of forbidden colors of e∈{uw1,uw2,vw1,vw2} is at most
1+(Δ−2)=Δ−1, |
which implies that the number of colors that can be used to color e is at least two. By the fact that each even cycle is 2-edge-choosable [2], we can color {uw1,uw2,vw1,vw2} properly.
Finally, we easily recolor the 2-vertices of this configuration and extend c to the whole graph G, a contradiction.
Lemma 3.3. Let uv∈E(G). If u∈S(G), then v∈B(G). (This lemma implies that two small vertices cannot be adjacent.)
Proof. Suppose to the contrary that both u∈S(G) and v∈S(G). By the choice of G, G−uv has a total-(Δ+1)-coloring c, then the number of forbidden colors of v is
|c(NG(v))∪c(NEG(v))|≤4ϵ(2−c′+ϵ)+[4ϵ(2−c′+ϵ)−1]=8ϵ(2−c′+ϵ)−1≤Δ−1. |
Therefore uv can receive any color except for c(u)∪c(v) and those colors in c(NEG(u)∪NEG(v)), then the number of forbidden colors of uv is
|c(u)∪c(v)∪NEG(u)∪NEG(v)|≤2+[4ϵ(2−c′+ϵ)−1]+[4ϵ(2−c′+ϵ)−1]=8ϵ(2−c′+ϵ)≤Δ. |
Thus, we can extend c to the whole graph G, a contradiction.
Lemma 3.4. There is no cut vertex in G.
Proof. Suppose to the contrary that there exists a cut vertex v in G with d(v)≤Δ. Let G1 be a v-component of G and G2=G[(V(G)∖V(G1))∪{v}]. Let NEG1(v)={vv1,vv2,...,vvs} and NEG2(v)={vvs+1,vvs+2,...,vvd(v)}. By the minimality of G, G1 has a total-(Δ+1)-coloring c1 and G2 has a total-(Δ+1)-coloring c2.
We notice that ∀vvi,vvj∈NEG1(v), c1(vvi)≠c1(vvj); ∀vvi,vvj∈NEG2(v), c2(vvi)≠c2(vvj). Let Xj⊆V(G1)∪E(G1) be the set that contains every vertex and edges that are colored by j in G1 for j∈[Δ+1], then for vvi∈{vv1,vv2,...,vvs}, we exchange the colors of Xc1(vvi) and Xi such that c(vvi)=i. We also exchange the colors of Xc1(v) and XΔ+1 such that c1(v)=Δ+1.
For G2, we repeat this procedure such that c(vvj)=j for vvj∈{vvs+1,vvs+2,...,vvd(v)} (Since G1 and G2 are colored separately, this procedure cannot change the colors in V(G1)). Note that d(v)≤Δ, which implies that there are at most Δ colors in c(NEG(v)). At last, we exchange the colors of Xc2(v) and XΔ+1 such that c2(v)=c1(v)=Δ+1.
Thus, we can get a total coloring c of G from c1 and c2, a contradiction.
Lemma 3.5. For each big vertex b1 of G, there exists another big vertex b2 such that the distance between b1 and b2 is at most two.
Proof. Suppose to the contrary that ∀b∈B(G) with b≠b1, dG(b1,b)≥3. Let P=b1v1v2 be a path, then vi∈S for i∈{1,2}, which implies that a small vertex v1 is adjacent to another small vertex v2, contradicting the fact that a small vertex can only be adjacent to a big vertex in Lemma 3.3.
Remark. In the following figures, the black solid vertices means big vertices, the white hollow vertices means small 2-vertices and the gray vertices means small 3+- vertices. Note that there is no 1-vertex in V(G) by Lemma 3.1.
Now we give some definitions and notations that we will use in the following sections.
Let P be a b1b2-path and |P|≤2. We say P is a link path of b1 and b2 in G if bi∈B(G) for i∈{1,2} and v∈S(G) for v∈V(P)∖{b1,b2}.
Let P be a link path of b1 and b2. We divide the link path P into three types according to the length and the type of its inner vertices. The sets of Type 1-3 link paths of G are denoted by L1-L3, respectively. Note that Lemmas 3.2, 3.3 and 3.5 ensure that the number of types of the link path is only three (As shown in Figure 3).
Type-1 link path | If P∈L1, then |P|=1. |
Type-2 link path | If P∈L2, then |P|=2 (let P=b1vb2) and d(v)=2. |
Type-3 link path | If P∈L3, then |P|=2 (let P=b1vb2) and d(v)≥3. |
As shown in Figure 4, let {b1,b2,...,bx} be the neighbors of a small 3+-vertex u in clockwise order, then bi∈B(G) for i∈{1,2,...,x}. For all bi∈{b1,b2,...,bx}, we say the link path biubi+1 is a special-Type-3 link path of bi (or bi+1), where x+1≡1(mod x). For example, b1ub2 is a special-Type-3 link path and b1ub3 is not a special-Type-3 link path.
Let b1,b2∈B(G). We say b1 is subadjacent to b2 if there exists a Type-1, Type-2 or special-Type-3 link path P of b1 and b2.
Assume that both P1 and P2 are link paths of b1 and b2. Let C∗=G[V(P1)∪V(P2)]. We assume that C∗ partitions the surface Σ into two disjoint arcwise-connected open sets. We say P1 is adjacent to P2 if V(int(C∗))=ϕ or V(ext(C∗))=ϕ.
Let P={P1,P2,…,Pm−1,Pm} be a sequence of the link paths of b1 and b2 and P is called an ordered sequence of the link paths of b1 and b2 if Pi is adjacent to Pi+1 for i∈{1,2,…,m−1}. We say P is a maximal ordered sequence of the link paths of b1 and b2 if, for any ordered sequence of the link paths P′={Px1,Px2,…,Pxn} that satisfy P⊆P′, we have P′=P.
According to the definitions and notations above, we give the following lemma about the link path.
Lemma 3.6. Let b1 and b2 be any two big vertices of G and P={P1,P2,…,Pr} be a maximal ordered sequence of the link paths of b1 and b2, then r≤4 (furthermore, if P only contains Type-3 link paths of b1 and b2, then r≤2).
Proof. As shown in Figure 5, the number of Type-1 link paths of b1 and b2 is at most one since G is a simple graph; the number of Type-2 link paths of b1 and b2 is at most one by Lemma 3.2.
Now we prove that the number of Type-3 link paths of b1 and b2 in P is at most two. Suppose to the contrary that there are three Type-3 link paths of b1 and b2 in P, which are denoted by {b1v1b2,b1v2b2,b1v3b2}.
Let C1=b1v1b2v2b1, C2=b1v2b2v3b1. By the definition of the maximal ordered sequence of the link paths, we know that both int(C1) and int(C2) are homeomorphic to discs. Furthermore, b1v2b2 is the common boundary of int(C1) and int(C2). Now, we can infer that the degree of v2 must be two since b1v2b2is adjacent to both b1v1b2 and b1v3b2, contradicting the fact that b1v2b2 is a Type-3 link path.
Therefore, r≤4.
In the following part, we will construct a simple graph G′ that can be embedded in Σ from G such that V(G′)=B(G) and δ(G′)≥ϵ+1, which contradicts the fact that every simple graph embedded in surface Σ has an ϵ−-vertex. Thus, the proof is completed.
Remark 1. We assume that G has been embedded in Σ and the position of each vertex in B(G) does not change in the following part.
Now, we are ready to construct a simple graph G′ from G such that V(G′)=B(G).
The construction process is listed below. Note that V(G0)=V(G1)=V(G2)=B(G) in the following part:
G0S1__G1S2__G2simplegraph__G′, |
First, we assume that G0 is an empty graph such that V(G0)=B(G), then we add edges to the empty graph G0 one by one by the following two steps.
The graph obtained after (S1) is denoted by G1
(S1). If there exists a Type-1 or Type-2 link path of b1 and b2 in G, we choose an arbitrary Type-1 or Type-2 link path P then add an edge b1b2 in G0. Make sure the edge b1b2 entirely overlaps with P.
Remark 2. It is obvious that G1 is a graph that can be embedded in Σ since each edge of G1 corresponds to a link path in graph G.
The graph that is obtained after (S1) and (S2) is denoted by G2. As shown in Figure 6, the real lines stand for the edges in G while the dotted lines stand for the edges in G2.
(S2). Let u′∈V(G)∖B(G) and d(u′)=x, NG(u′)={b1,b2,…,bx} in clockwise order and note that bi∈B(G) for i∈{1,2,…,x}. For any i∈{1,2,…,x}, let P′i=biu′bi+1 where x+1≡1(mod x), then P′i is a link path of bi and bi+1 in G. Add an edge bibi+1 in G1 such that the edge bibi+1 is as close as possible to P′i (only close, not overlapping).
Remark 3. It is obvious that the edges that are added in (S2) only intersect at their ends, and each edge that is added in (S2) only intersects with the edges that are added in (S1) at their ends. Thus, we know that G2 can be embedded in Σ.
Let G′ be the simple graph of G2 (with no loops or parallel edges). Next we will prove that δ(G′)≥ϵ+1 and get a contradiction.
Claim 3.1. If there exists a path P such that P is a Type-1 or Type-2 link path of b1 and b2 in G, then b1 is adjacent to b2 in G′.
Proof. It is easy to verify this claim by (S1).
Claim 3.2. If there exists a path P such that P is a Type-3 link path of b1 and b2 in G, let P=b1u′b2, then there exist two vertices b,b∗∈NG(u′) such that {b,b∗}⊆NG′(b1).
Proof. It is easy to verify this claim by (S2). (Note that b1u′b and biu′b∗ are special-Type-3 link paths. For example, as shown in Figure 6, {b2,bx}⊆NG′(b1).)
Lemma 3.7. δ(G′)≥ϵ+1.
Proof. Note that V(G′)=B(G). Let b be an arbitrary vertex in V(G′). Let bi∈B(G). Recall that b is subadjacent to bi in G if there exists a Type-1, Type-2 or special-Type-3 link path P of b and bi in G.
Case 1. If b is subadjacent to at least ϵ+1 big vertices in G, it is easy to verify that dG′(b)≥ϵ+1 by Claims 3.1 and 3.2.
Case 2. If b is subadjacent to at most t (t≤ϵ) big vertices {b1,b2,...,bt} in G, then there exists b′∈{b1,b2,...,bt} such that the number of link paths of b and b′ in G is at least ⌈4ϵ(2−c′+ϵ)+1t⌉≥4(2−c′+ϵ) by pigeonhole principle. The number of maximal ordered sequence of the link paths of b and b′ is at least (2−c′+ϵ) by Lemma 3.6. We assume that the number of maximal ordered sequences of the link paths of b and b′ is (2−c′+ϵ), otherwise we can get an easier proof.
Let P1,P2,...,P2−c′+ϵ be the maximal ordered sequence of the link paths of b and b′, then ∀Pi∈{P1,P2,...,P2−c′+ϵ}. We regard Pi as a new path P′i as shown in Figure 7 (the gray region is a new path of bb′), then (2−c′+ϵ) paths P′1,P′2,...,P′2−c′+ϵ can form (2−c′+ϵ−1) simple closed curves C1,C2,...,C2−c′+ϵ−1 in Σ. By Corollary 3, we know that (2−c′+ϵ−1) simple closed curves partition Σ into at least ϵ disjoint arcwise-connected open sets. Without loss of generality, let Σ∖(⋃2−c′+ϵ−1i=1Ci)=D1∪D2∪...∪Dϵ, where Di is the arcwise-connected open set and they disjoint with each other for Di∈{D1,D2,...,Dϵ}. In the following part, we prove that b is adjacent to at least ϵ+1 vertices in G′.
Subcase 2.1. If ∀Di∈{D1,D2,...,Dϵ}, there exists a Type-3 link path Pi of b and b′ such that Pi is on the boundary of Di. Let Pi=bub′, then u∈S(G) and d(u)≥3. It is obvious that there is neither a b-component nor a b′-component in Di by Lemma 3.4 (as shown in Figure 8(1), the white line means a vertex with uncertain degree). Recall that two small vertices cannot be adjacent by Lemma 3.3, which implies that there exists bi∈Di such that bi∈NG(u) and bi is subadjacent to b in G (as shown in Figure 8(2)). It is easy to verify that bi is adjacent to b in G′ by Claim 3.2.
Therefore, for all Di∈{D1,D2,...,Dϵ}, there exists bi∈Di (bi≠b′) such that bi is a big vertex and bi is adjacent to b in G′, and it is obvious that b′ is adjacent to b in G′. Thus, we know that dG′(b)≥ϵ+1.
Subcase 2.2. If there exists Di∈{D1,D2,...,Dϵ} such that the boundary of Di contains either Type-1 or Type-2 link paths of b and b′, note that the number of Type-1 (or Type-2) link paths of b and b′ is at most one by Lemma 3.2, which implies that the boundary of Di contains exactly one Type-1 link path and one Type-2 link path of b and b′. Recall that the number of link paths of b and b′ in G is at least 4(2−c′+ϵ), then the number of maximal ordered sequence of the link paths of b and b′ is at least ⌈4(2−c′+ϵ)−1−12⌉≥(2−c′+ϵ)+1.
Now (2−c′+ϵ)+1 maximal ordered sequence of the link paths of b and b′ can form (2−c′+ϵ)+1 paths, (2−c′+ϵ)+1 paths can form (2−c′+ϵ) simple closed curves and (2−c′+ϵ) simple closed curves partition Σ into at least ϵ+1 disjoint arcwise-connected open sets {D′1,D′2,...,D′ϵ+1}, then ∀D′j∈{D′1,D′2,...,D′ϵ+1}∖{Di}, we can find a Type-3 link path P′j of b and b′ such that P′j is on the boundary of D′j (since D′j≠Di). Let P′j=bu′b′, then u′∈S(G) and d(u′)≥3, which implies that there exists b′j∈D′j such that b′j∈NG(u′) and b′j is subadjacent to b. It is easy to verify that b′j is adjacent to b in G′ by Claim 3.2.
Therefore, for all ∀D′j∈{D′1,D′2,...,D′ϵ+1}∖{Di}, there exists b′j∈D′j (b′j≠b′) such that b′j is a big vertex and b′j is adjacent to b in G′, and it is obvious that b′ is adjacent to b in G′. Thus, we know that dG′(b)≥ϵ+1.
Now, we get a simple graph G′ that can be embedded in Σ such that V(G′)=B(G) and δ(G′)≥ϵ+1, which contradicts the fact that every simple graph embedded in Σ has an ϵ−-vertex. The proof is completed.
The authors declare they have not used Artificial Intelligence (AI) tools in the creation of this article.
Supported by the Natural Science Foundation of China, Grant No.11871377.
The authors declare that they have no conflicts of interest regarding the publication of this paper.
[1] | M. Armstrong, Basic Topology, Berlin: Springer-Verlag, 1983. https://doi.org/10.1007/978-1-4757-1793-8 |
[2] | J. Bondy, U. Murty, Graph Theory, Berlin: Springer Press Graduate Texts in Mathematics (GTM244), 2008. https://doi.org/doi:10.1007/978-1-84628-970-5 |
[3] | M. Behzad, Graphs and their chromatic numbers, Ph.D thesis, Michigan State University, 1965. https://doi.org/doi:10.25335/M5H41K22C |
[4] | O.V. Borodin, Coupled colorings of graphs on a plane, Metody Diskret. Analiz, in Russian, 45 (1987), 21–27. |
[5] |
O. V. Borodin, On the total coloring of planar graphs, J. Reine Angew. Math., 394 (1989), 180–185. https://doi.org/10.1515/crll.1989.394.180 doi: 10.1515/crll.1989.394.180
![]() |
[6] |
O. V. Borodin, A. V. Kostochka, D. R. Woodall, List edge and list total colourings of multigraphs, J. Combin. Theory Ser. B, 71 (1997), 184–204. https://doi.org/10.1006/jctb.1997.1780 doi: 10.1006/jctb.1997.1780
![]() |
[7] |
O. V. Borodin, A. V. Kostochka, D. R. Woodall, Total colourings of planar graphs with large maximal degree, J. Graph Theory, 26 (1997), 53–59. https://doi.org/10.1002/(SICI)1097-0118(199709)26:1<53::AID-JGT6>3.0.CO;2-G doi: 10.1002/(SICI)1097-0118(199709)26:1<53::AID-JGT6>3.0.CO;2-G
![]() |
[8] | T. R. Jensen, B. Toft, Graph Coloring Problems, Hoboken: Wiley Interscience, 1995. |
[9] |
A. V. Kostochka, The total coloring of a multigraph with maximal degree 4, Discrete Math., 17 (1977), 161–163. https://doi.org/10.1016/0012-365X(77)90146-7 doi: 10.1016/0012-365X(77)90146-7
![]() |
[10] | A. V. Kostochka, An analogue of Shannon's estimate for complete colorings, Metody Diskret. Analiz, in Russian, 30 (1977), 13–22. |
[11] |
A. V. Kostochka, The total chromatic number of any multigraph with maximum degree five is at most seven, Discrete Math., 162 (1996), 199–214. https://doi.org/10.1016/0012-365X(95)00286-6 doi: 10.1016/0012-365X(95)00286-6
![]() |
[12] |
L. Kowalik, J. S. Sereni, R. Škrekovski, Total colouring of plane graphs with maximum degree nine, SIAM J. Discrete Math., 4 (2008), 1462–1479. https://doi.org/10.1137/070688389 doi: 10.1137/070688389
![]() |
[13] | B. Mohar, C. Thomassen, Graphs on Surfaces, New York: Springer, 2001. https://doi.org/10.1007/978-1-4614-6971-1 |
[14] |
D. P. Sanders, Y. Zhao, On total 9-coloring planar graphs of maximum degree seven, J. Graph Theory, 31 (1999), 67–73. https://doi.org/10.1002/(SICI)1097-0118(199901)30:1<67::AID-JGT7>3.0.CO;2-M doi: 10.1002/(SICI)1097-0118(199901)30:1<67::AID-JGT7>3.0.CO;2-M
![]() |
[15] | V. G. Vizing, Some unsolved problems in graph theory, Uspekhi Mat. Nauk, in Russian, 23 (1968), 117–134. https://doi.org/10.1070/rm1968v023n06abeh001252 |
[16] |
W. Wang, Total chromatic number of planar graphs with maximum degree ten, J. Graph Theory, 54 (2014), 91–102. https://doi.org/10.1002/jgt.20195 doi: 10.1002/jgt.20195
![]() |
[17] |
H. Wang, B. Liu, J. Wu, Total coloring of graphs embedded in surfaces of nonnegative Euler characteristic, Sci. China Math., 57 (2014), 211–220. https://doi.org/10.1007/s11425-013-4576-2 doi: 10.1007/s11425-013-4576-2
![]() |