Research article

AVD edge-colorings of cubic Halin graphs

  • Received: 25 June 2023 Revised: 11 September 2023 Accepted: 18 September 2023 Published: 09 October 2023
  • MSC : 05C15

  • The adjacent vertex-distinguishing edge-coloring of a graph G is a proper edge-coloring of G such that each pair of adjacent vetices receives a distinct set of colors. The minimum number of colors required in an adjacent vertex-distinguishing edge-coloring of G is called the adjacent vertex-distinguishing chromatic index. In this paper, we determine the adjacent vertex distinguishing chromatic indices of cubic Halin graphs whose characteristic trees are caterpillars.

    Citation: Ningge Huang, Lily Chen. AVD edge-colorings of cubic Halin graphs[J]. AIMS Mathematics, 2023, 8(11): 27820-27839. doi: 10.3934/math.20231423

    Related Papers:

    [1] Baolin Ma, Chao Yang . Distinguishing colorings of graphs and their subgraphs. AIMS Mathematics, 2023, 8(11): 26561-26573. doi: 10.3934/math.20231357
    [2] 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
    [3] Shuangliang Tian, Ping Chen . Edge-coloring of generalized lexicographic product of graphs. AIMS Mathematics, 2024, 9(6): 15988-15995. doi: 10.3934/math.2024774
    [4] 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
    [5] Meiqin Jin, Ping Chen, Shuangliang Tian . Interval edge colorings of the generalized lexicographic product of some graphs. AIMS Mathematics, 2024, 9(11): 30597-30611. doi: 10.3934/math.20241477
    [6] 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
    [7] Yunfeng Tang, Huixin Yin, Miaomiao Han . Star edge coloring of $ K_{2, t} $-free planar graphs. AIMS Mathematics, 2023, 8(6): 13154-13161. doi: 10.3934/math.2023664
    [8] Shabbar Naqvi, Muhammad Salman, Muhammad Ehtisham, Muhammad Fazil, Masood Ur Rehman . On the neighbor-distinguishing in generalized Petersen graphs. AIMS Mathematics, 2021, 6(12): 13734-13745. doi: 10.3934/math.2021797
    [9] Jin Cai, Shuangliang Tian, Lizhen Peng . On star and acyclic coloring of generalized lexicographic product of graphs. AIMS Mathematics, 2022, 7(8): 14270-14281. doi: 10.3934/math.2022786
    [10] Jahfar T K, Chithra A V . Central vertex join and central edge join of two graphs. AIMS Mathematics, 2020, 5(6): 7214-7233. doi: 10.3934/math.2020461
  • The adjacent vertex-distinguishing edge-coloring of a graph G is a proper edge-coloring of G such that each pair of adjacent vetices receives a distinct set of colors. The minimum number of colors required in an adjacent vertex-distinguishing edge-coloring of G is called the adjacent vertex-distinguishing chromatic index. In this paper, we determine the adjacent vertex distinguishing chromatic indices of cubic Halin graphs whose characteristic trees are caterpillars.



    All graphs considered in this paper are simple, finite and undirected. Let G=(V,E) be a graph with maximum degree Δ and c:E{1,2,,k} be an edge-coloring of G. For each vertex vV, the neighborhood N(v) of v is N(v)={u:uV,uvE}, we define the palette of v as S(v)={c(uv):uN(v)}, and denote by Sc(v) the complementary set of S(v) in {1,2,,k}. We call c a proper edge-coloring if it assigns distinct colors to adjacent edges. The minimum number of colors needed in a proper edge-coloring is the chromatic index of G, denoted by χ(G). An adjacent vertex-distinguishing edge-coloring (AVD edge-coloring for short) of G is a proper edge-coloring c such that S(v)S(u) for each uvE. The smallest integer k such that G has an AVD edge-coloring with k colors is called the adjacent vertex-distinguishing chromatic index (AVD chormatic index for short), denoted by χavd(G). Note that G has an AVD edge-coloring if and only if G has no isolated edges, we call this graph a normal graph. From the definition, for a normal graph G, we have χavd(G)χ(G)Δ, and if G contains two adjacent vertices of maximum degree, then χavd(G)Δ+1.

    The concept of AVD edge-coloring was first introduced by Zhang et al. [1], they completely determined χavd(G) for some special graphs such as paths, cycles, trees, complete graphs, and complete bipartite graphs, and proposed the following conjecture.

    Conjecture 1.1. [1] If G is a normal connected graph with |V(G)|3 and GC5. Then χavd(G)Δ(G)+2.

    Balister et al. [2] comfirmed Conjecture 1.1 for all graphs with maximum degree 3.

    Theorem 1.1. [2] If G is a graph with no isolated edges and Δ=3, then χavd(G)5.

    They showed that Conjecture 1.1 is also true for bipartite graphs, and if the chromatic number of G is k, then χavd(G)Δ(G)+O(logk). By using probabilistic method, Hatami [3] proved that χavd(G)Δ(G)+300 for graphs G with maximum degree Δ1020. Joret et al. [4] reduced this bound to Δ+19. Horňák et al. [5] showed that Conjecture 1.1 holds for planar graphs with maximum degree at least 12. Yu et al. [6] verified this conjecture for graphs with maximum degree at least 5 and maximum average degree less than 3. In addition, there are many graphs with adjacent vertex-distinguishing chromatic indices at most Δ(G)+1. Hocquard and Montassier[7] showed that χavd(G)Δ(G)+1 for graphs with Δ(G)5 and mad(G)<22Δ(G). Bonamy and Przybyło [8] proved that for any planar graph G with Δ(G)28 and no isolated edges, χavd(G)Δ(G)+1. Huang et al. [9] showed that χavd(G)Δ(G)+1 holds for every connected planar graph G without 3-cycles and with maximum degree at least 12.

    Wang et al. [10] proved that χavd(G)max{6,Δ(G)+1} for any 2-degenerate graph G without isolated edges. Wang and Wang [11] characterized the adjacent vertex-distinguishing chromatic indices for K4-minor graphs. Cubic Halin graphs is an important class of graphs, Chang and Liu [12] considered the strong edge-coloring of cubic Halin graphs. In this paper, we will study the adjacent vertex-distinguishing edge-coloring of cubic Halin graphs.

    A Halin graph G is a plane embedding of a tree T and a cycle C, where the inner vertices of T have minimum degree at least 3, and the cycle C connects all the leaves of T in such a way that C is the boundary of the exterior face. The tree T and the cycle C are called the characteristic tree and the adjoint cycle of G, respectively.

    A caterpillar is a tree whose removal of leaves results in a path P (called spine of the caterpillar). Let Gr be the set of all cubic Halin graphs whose characteristic trees are caterpillars with r+2 leaves. For a Halin graph G=TC in Gr, denote the spine P of T as P=v1v2vr, let u0, u1 be the neighbors of v1 other than v2, and ur, ur+1 be the neighbors of vr other than vr1. For 2ir1, let ui be the neighbor of vi that is a leaf of T. Moreover, assume that u1u2E(G) and ur1urE(G). Let v be a vertex of P. We call u a leaf-neighbor of v if u is adjacent to v and is of degree 1 in T, and the edge uv is called the leaf-edge. We draw G on the plane by putting the spine P vertically in the middle, and the leaf-edges incident with vi, 2ir1, either left or right edges horizontally to P. See Figure 1 for an example of G8.

    Figure 1.  The graph H0.

    In particular, if all the leaf-neighbors are on the same side of P, then we call this graph G a necklace and denote by Nr. We give configurations of N4 and N5 in Figure 2.

    Figure 2.  The necklace N4 and N5.

    It's easy to see that χavd(G)4 for any cubic graph G. By Theorem 1.1, the adjacent vertex-distinguishing chromatic index for any cubic Halin graph is at most 5. Hence, for any cubic Halin graph G, we have either χavd(G)=4 or χavd(G)=5. Thus it is interesting to determine the exact value of χavd(G). In this paper, we consider the cubic Halin graphs in Gr, and show that there are only two graphs in Gr with the AVD chromatic index 5.

    Theorem 1.2. Let r2 be an integer and GGr. Then χavd(G)=4 if G{N4,N5}; otherwise χavd(G)=5.

    Let G be a cubic Halin graph in Gr. We define the subgraphs induced by {u1v1,u0u1,u0v1,u1u2,v1v2,u0ux} and {urur+1,urvr,vrur+1,ur1ur,vr1vr,uyur+1} as end-graphs of G, where ux and uy are the neighbors of u0 and ur+1, see Figure 3 for an illustration. We denote these two subgraphs by G1 and Gr, respectively. For a vertex ui (2ir1), we will use ui and ui to denote the neighbors of ui that are on the cycle if the neighbors of ui are uncertain, where ui is closer to the end-graph G1. For 2ir1, if the leaf-neighbors of vi,vi+1,,vi+k1 are on the same side of P, while vi1 and vi+k have leaf-neighbors on the other side. Then the subgraph induced by {vi,vi+1,,vi+k1,ui,ui+1,,ui+k1} accompanied by the extra edges vi+k1vi+k and ui+k1ui+k1 is called a k-block, denoted by Gi,k. If a k-block contains the vertex vr, then the block is a bottom block of G. See the graph H0 in Figure 1, the subgraph induced by {u6,v6,u7,v7} accompanied by the edges u7u8 and v7v8 is the 2-block G6,2 and it is a bottom block. For two blocks Gi,k and Gj,t, if vi+k=vj or vj+t=vi, then we say Gi,k and Gj,t are adjacent. If vi+kvj, then we say Gi,k is before Gj,t. We call a subgraph obtained from the union of k adjacent 1-block a k-crossing block, or crossing block for short, of G. We denote the k-crossing block obtained from the union of Gi,1,Gi+1,1,,Gi+k1,1 as Gi,k,c. In Figure 1, the graph induced by the edges {v4u4,v4v5,u4u6,v5u5,v5v6,u5u9} is the 2-crossing block G4,2,c.

    Figure 3.  The end graphs G1 and Gr.

    A coloring of G is good, if it is an AVD-edge-coloring of G using colors in {1,2,3,4}. To prove Theorem 1.2, we will give a good coloring of G by coloring the edges of G from the top down. Initially we establish a good coloring of the end-graph G1, then we extend this coloring to the block that contains u2v2. By analyzing the coloring of G1 and the block containing u2v2, we proceed to color the block that is adjacent to the block containing u2v2. Repeat this process until we complete the coloring of the bottom block and the end-graph Gr.

    In the following, given an edge-coloring c of a graph G, we define a vertex coloring ¯c respect to c as follows: for each vertex vV(G), let ¯c(v) be an element in Sc(v), that is, ¯c(v) is the color that is not appeared at the edges incident with v. Note that if c is a good coloring of G, uvE(G), and d(u)=d(v)=3, then ¯c(u) and ¯c(v) are unique, and ¯c(u)¯c(v). Now we consider the colorings of the end-graphs.

    Proposition 2.1. Let G1 be an end-graph with vertex set {u1,u2,v1,v2,u0,ux}. If G1 admits a good coloring, then at least two edges of u1u2,v1v2, and u0ux are colored the same. Moreover, there are four types of good colorings of G1:

    (1) c(u1u2)=c(v1v2)=c(u0ux), c(u1v1)=¯c(u0), c(u0v1)=¯c(u1),c(u0u1)=¯c(v1);

    (2) c(u1u2)=c(v1v2)c(u0ux), c(u0v1)=¯c(u1), c(u0u1)=¯c(v1),c(u1u2)=¯c(u0);

    (3) c(u1u2)=c(u0ux)c(v1v2), c(u1v1)=¯c(u0), c(u0v1)=¯c(u1),c(u1u2)=¯c(v1);

    (4) c(v1v2)=c(u0ux)c(u1u2), c(u1v1)=¯c(u0), c(u0u1)=¯c(v1),c(v1v2)=¯c(u1).

    Proof. Suppose that G1 has a good coloring ϕ. If u1u2,v1v2, and u0ux are colored with distinct colors, without loss of generality, assume that ϕ(u1u2)=1,ϕ(v1v2)=2 and ϕ(u0ux)=3, then ϕ(u0u1){2,4}. If ϕ(u0u1)=4, then ϕ(u1v1)=3 and ϕ(u0v1)=1. But then S(u0)=S(u1), contradicts that ϕ is a good coloring. Hence ϕ(u0u1)=2. No matter what color of u1v1 is, we always have S(u1)=S(v1), so ϕ cannot be a good coloring. Therefore, at least two edges of u1u2,v1v2, and u0ux are colored the same.

    There are four types of colorings on the edges u1u2,v1v2 and u0ux such that at least two of them are colored the same, each type we will obtain a good coloring of G1. Let c be a coloring of the edges u1u2,v1v2 and u0ux.

    Type (1): c(u1u2)=c(v1v2)=c(u0ux). Then color u0u1,u0v1,v1u1 with three distinct colors in {1,2,3,4} that are different from c(u1u2). Thus, c is a good coloring of G1, and c(u1v1)=¯c(u0), c(u0v1)=¯c(u1),c(u0u1)=¯c(v1).

    Type (2): c(u1u2)=c(v1v2)c(u0ux). Then color u0u1, u0v1 with distinct colors in {1,2,3,4}{c(u1u2),c(u0ux)}, and color u1v1 with c(u0ux). Thus, c is a good coloring of G1, and c(u0v1)=¯c(u1), c(u0u1)=¯c(v1),c(u1u2)=¯c(u0).

    Type (3): c(u1u2)=c(u0ux)c(v1v2). Then color u1v1, u0v1 with distinct colors in {1,2,3,4}{c(v1v2),c(u0ux)}, and color u0u1 with c(v1v2). Thus, c is a good coloring of G1, and c(u1v1)=¯c(u0), c(u0v1)=¯c(u1),c(u1u2)=¯c(v1).

    Type (4): c(v1v2)=c(u0ux)c(u1u2). Then color u0u1, u1v1 distinct colors in {1,2,3,4}{c(u1u2),c(v1v2)}, and color u0v1 is with c(u1u2). Thus, c is a good coloring of G1, and c(u1v1)=¯c(u0), c(u0u1)=¯c(v1),c(v1v2)=¯c(u1).

    Therefore, we complete the proof of this proposition.

    Remark 2.1. The results of Proposition 2.1 is also holds for the end-graph Gr, that is, if Gr admits a good coloring, then at least two edges of ur1ur,vr1vr,uyur+1 are colored the same.

    Lemma 2.1. Let G be a graph in Gr, and G be the graph obtained from G by deleting edges urvr,vrur+1, and urur+1. Suppose that G has a good coloring c, then c can be extended to a good coloring of G if and only if one of the following statements holds:

    (1) c(ur1ur)=c(vr1vr)=c(uyur+1);

    (2) c(ur1ur)=c(vr1vr)c(uyur+1) and ¯c(uy)c(ur1ur);

    (3) c(ur1ur)=c(uyur+1)c(vr1vr) and ¯c(vr1)c(ur1ur), moreover, if ¯c(ur1)=¯c(uy), then they are equal to c(vr1vr);

    (4) c(vr1vr)=c(uyur+1)c(ur1ur) and ¯c(ur1)c(vr1vr), moreover, if ¯c(vr1)=¯c(uy), then they are equal to c(ur1vr).

    Proof. Suppose c is extended to a good coloring of G, then c is a good coloring of Gr. By Remark 2.1, at least two edges of ur1ur,vr1vr, and uyur+1 are colored the same. If all three edges ur1ur,vr1vr, and uyur+1 are colored the same, then statement (1) holds. Otherwise, exactly two edges of them are colored the same.

    If c(ur1ur)=c(vr1vr)c(uyur+1), then c(vrur+1)c(ur1ur) since c(vrur+1)c(vr1vr). Furthermore, c(urur+1)c(ur1ur) and c(uyur+1)c(ur1ur), hence c(ur1ur) does not appear at the edges incident with ur+1, it follows that ¯c(ur+1)=c(ur1ur). Because ¯c(uy)¯c(ur+1), we have ¯c(uy)c(ur1ur).

    If c(ur1ur)=c(uyur+1)c(vr1vr), without loss of generality, assume that c(ur1ur)=c(uyur+1)=1, c(vr1vr)=2, then c(urvr)1 and c(vrur+1)1, hence ¯c(vr)=1, which implies that ¯c(vr1)1, that is, ¯c(vr1)c(ur1ur). Furthermore, if ¯c(ur1)=¯c(uy), then ¯c(ur1) must appear on urur+1. If ¯c(ur1)c(vr1vr), then ¯c(ur1){3,4}. If ¯c(ur1)=3, then c(urur+1)=3, so c(urvr)=4, and vrur+1 cannot be colored. If ¯c(ur1)=4, then c(urur+1)=4, so c(urvr)=3, and vrur+1 cannot be colored. Therefore, ¯c(ur1)=c(vr1vr).

    If c(vr1vr)=c(uyur+1)c(ur1ur), by the same analysis as case c(ur1ur)=c(uyur+1)c(vr1vr), we have ¯c(ur1)c(vr1vr), and if ¯c(vr1)=¯c(uy), then they must equal to c(ur1vr).

    Therefore, if c is extended to a good coloring of G, then one of the statements (1)–(4) holds.

    On the other hand, we show that if c satisfies one of the statements (1)–(4), then c can be extended to a good coloring of G.

    Suppose that c satisfies statment (1), that is, c(ur1ur)=c(vr1vr)=c(uyur+1). Since ur1vr1E(G), we have ¯c(ur1)¯c(vr1). If ¯c(uy) is distinct from ¯c(ur1) and ¯c(vr1), then let c(urvr)=¯c(ur1),c(vrur+1)=¯c(vr1),c(urur+1)=¯c(uy). It is easy to see that c is a good coloring of G. If ¯c(uy) is equal to ¯c(ur1) or ¯c(vr1), without loss of generality, assume that ¯c(uy)=¯c(ur1), then let c(urur+1)=¯c(uy),c(urvr)=¯c(vr1),c(vrur+1)={1,2,3,4}{¯c(uy),¯c(vr1),c(vr1vr)}. Then c is a good coloring of G.

    Now suppose that c satisfies statment (2), that is, c(ur1ur)=c(vr1vr)c(uyur+1). Without loss of generality, assume that c(ur1ur)=c(vr1vr)=1 and c(uyur+1)=2. By statement (2), ¯c(uy)1. If ¯c(ur1)=2, then let c(urvr)=2,c(vrur+1)=¯c(vr1), and c(urur+1)={1,2,3,4}{1,2,¯c(vr1)}. If ¯c(vr1)=2, then let c(urvr)=2,c(urur+1)=¯c(ur1), and c(vrur+1)={1,2,3,4}{1,2,¯c(ur1)}. If ¯c(ur1)2 and ¯c(vr1)2, then let c(urvr)=2,c(urur+1)=¯c(ur1), and c(vrur+1)=¯c(vr1). Note that ¯c(ur1)1, ¯c(vr1)1, and ¯c(ur1)¯c(vr1), hence all the colorings above are good colorings of G.

    Next suppose that c satisfies statment (3), that is, c(ur1ur)=c(uyur+1)c(vr1vr). Without loss of generality, assume that c(ur1ur)=c(uyur+1)=1 and c(vr1vr)=2. If ¯c(ur1)=¯c(uy), by statement (3), ¯c(ur1)=¯c(uy)=2, then let c(urur+1)=2,c(urvr)=¯c(vr1), and c(vrur+1)={1,2,3,4}{1,2,¯c(vr1)}. If ¯c(ur1)=2,¯c(uy)2, then let c(urur+1)=2,c(urvr)=¯c(vr1), and c(vrur+1)={1,2,3,4}{1,2,¯c(vr1)}. If ¯c(ur1)2, then let c(urvr)=¯c(ur1), c(vrur+1)=¯c(vr1) and c(urur+1){1,2,3,4}{1,¯c(ur1),¯c(vr1)}. Note that ¯c(vr1)2, and by statement (3), we have ¯c(vr1)1, hence ¯c(vr1){3,4}. Therefore, we can check that the colorings above are good colorings of G.

    The argument for statement (4) is similar as the argument for statement (3), hence we omit the proof here.

    Next we consider the coloring of the blocks. Let Gi,k (Gi,k,c) be a k-block (k-crossing block), we define the associated subgraph Hi,k (Hi,k,c) of Gi,k (Gi,k,c) as the subgraph obtained by the union of G1 and all the blocks before Gi,k (Gi,k,c). To color Gi,k or Gi,k,c, we assume that the associated subgraph Hi,k has a good coloring c. Let vjuj be an edge with ji. We define {c(vj1vj),c(ujuj),¯c(vj1),¯c(uj)} as the total-set of vjuj. If the total-set of vjuj is {1,2,3,4}, then we call vjuj a full-edge. If c(vj1vj)c(ujuj),c(vj1vj)=¯c(uj),c(ujuj)¯c(vj1), then we call vjuj an in-half-edge. If c(vj1vj)c(ujuj),c(ujuj)=¯c(vj1),c(vj1vj)¯c(uj), then we call vjuj an out-half-edge. A half-edge means a in-half-edge or out-half-edge. The edge vjuj is a crossing-edge if c(ujuj)=c(vj1vj). Note that, if vjuj is a crossing-edge, then c(ujuj)=¯c(vj) and c(vjvj+1)=¯c(uj), and vice versa. For two edges vjuj and vj+1uj+1, assume that uj and uj+1 are on the different sides of P, we call vjuj an outer-crossing-edge if c(vjvj+1)=¯c(uj) and c(ujuj)=¯c(uj+1). If c(ujuj){¯c(vj),¯c(uj+1)}, then we call the color c(ujuj) suitable. Note that if vjuj is a crossing-edge or outer-crossing-edge, then c(ujuj) is suitable.

    Lemma 2.2. Let Gi,k be a k-block with k2, suppose Hi,kGi,k has a good coloring c such that vjuj is an in-half-edge(out-half-edge) for some j, ij<i+k1, then for any t, j<ti+k1, vtut is also an in-half-edge(out-half-edge).

    Moreover, if Gi,k is a bottom block and vjuj is a half-edge, then c can be extended to a good coloring of G if and only if c(ui1ur+1)=c(ur1ur) when vjuj is an in-half-edge or c(ui1ur+1)=c(vr1vr) when vjuj is an out-half-edge.

    Proof. We assume that vjuj is an in-half-edge. Without loss of generality, suppose c(vj1vj)=¯c(uj)=1, c(ujuj)=2, and ¯c(vj1)=3. Since ¯c(uj) must appear at the edges incident with uj and c(vjuj)1, we have that c(ujuj+1)=1. If c(vjvj+1)=2, then S(vj)=S(uj), contradicts that c is good coloring. So c(vjvj+1){3,4}. If c(vjvj+1)=3, then c(vjuj)=4. If c(vjvj+1)=4, then c(vjuj)=3. No matter vjvj+1 is colored with 3 or 4, we have c(vjvj+1)c(ujuj+1), c(vjvj+1)=¯c(uj), and c(ujuj+1)¯c(vj). That is, the edge vj+1uj+1 is a in-half-edge. By the same argument, we have that for any t, j<ti+k1, vtut is an in-half-edge. Furthermore, if Gi,k is a bottom block, then c(vr1vr)c(ur1ur), c(vr1vr)=¯c(ur1), and c(ur1ur)¯c(vr1). Statement (1), (2) and (4) of Lemma 2.1 can not hold. Hence c can be extended to a good coloring of G if and only if statement (3) of Lemma 2.1 holds. Since c(ur1ur)¯c(vr1), we have c(ui1ur+1)=c(ur1ur).

    By the same argument as above, we can show that if vjuj is an out-half-edge, then for any t, j<ti+k1, vtut is also an out-half-edge. And if Gi,k is a bottom block, then c can be extended to a good coloring of G if and only if c(ui1ur+1)=c(vr1vr).

    Lemma 2.3. Let Gi,k be a k-block with k4. Suppose Hi,k has a good coloring c such that viui is an in-half-edge (out-half-edge), then for any α{1,2,3,4}, c can be extended to a good coloring of Gi,k such that c(ui+k1ui+k1)=α (c(vi+k1vi+k)=α).

    Proof. Suppose viui is an in-half-edge, without loss of generality, assume that c(vi1vi)=¯c(ui)=1,c(uiui)=2, and ¯c(vi1)=3. Then we have c(uiui+1)=1, and c(vivi+1){1,2}. We color vivi+1 with a color in {3,4} and color vi+k2vi+k1 with α. For i+1ji+k3, we color vjvj+1 with a color in {1,2,3,4} that is different from the colors of vj2vj1, vj1vj and α, and color vi+k1vi+k with a color different from the colors of vi+k3vi+k2 and vi+k2vi+k1. Then set c(ujuj+1)=c(vj1vj) for iji+k2 and c(ui+k1ui+k1)=α. Finally, for iji+k1, set c(vjuj)={1,2,3,4}{c(vj1vj),c(ujuj),c(vjvj+1}. It is easy to see that this coloring c is a good coloring of Gi,k and c(ui+k1ui+k1)=α.

    By symmetry, if viui is an out-half-edge, then for any α{1,2,3,4}, c can be extended to a good coloring of Gi,k such that c(vi+k1vi+k)=α.

    Lemma 2.4. Let Gi,k be a bottom block with k1. If Hi,k has a good coloring c such that viui is a crossing-edge, then c cannot be extended to a good coloring of G.

    Proof. First assume that k=1, then i=r1. If vr1ur1 is a crossing-edge, then c(ur1ur)=¯c(vr1) and c(vr1vr)=¯c(ur1). Hence statement (3) and (4) of Lemma 2.1 can not hold. Since ¯c(ur1)c(ur1ur), we have c(ur1ur)c(vr1vr). It follows that statement (1) and (2) of Lemma 2.1 can not hold. Therefore, by Lemma 2.1, c cannot be extended to a good coloring of G.

    Suppose k2. If viui is a crossing-edge, then c(uiui+1)=¯c(vi) and c(vivi+1)=¯c(ui). Without loss of generality, assume that c(uiui+1)=¯c(vi)=1 and c(vivi+1)=¯c(ui)=2, then ui+1ui+1 must be colored with 2 and vi+1vi+2 must be colored with 1. But then S(ui+1)=S(vi+1) no matter what color of ui+1vi+1 is, which shows that c cannot be extended to a good coloring of G.

    Theorem 2.1. If G is a necklace in Gr for r2, then χavd(G)=4 if r{4,5}, otherwise χavd(G)=5.

    Proof. If r=2, then G is the graph depicted in Figure 4. Let c(u0v1)=c(u2v2)=1,c(u0u1)=c(v2u3)=2,c(u1v1)=c(u2u3)=3, and c(u0u3)=c(v1v2)=c(u1u2)=4. It is easy to check that c is a good coloring of G.

    Figure 4.  The necklace N2.

    Now we assume that r3. Note that G is the union of end-graphs G1, Gr, and a (r2)-bottom block. We first give a good coloring of G1. By Proposition 2.1, there are four types of colorings on G1. In type (1) and (2), c(u1u2)=c(v1v2), which means u2v2 is a crossing edge, by Lemma 2.4, this coloring cannot be extended to a good coloring of G.

    In type (3), c(u1u2)=c(u0ur+1), c(u1u2)c(v1v2), c(u1u2)=¯c(v1), and c(u0v1)=¯c(u1). Since c(u0v1)c(v1v2), we have c(v1v2)¯c(u1), which means that v2u2 is an out-half-edge. By Lemma 2.2, c can be extended to a good coloring of G if and only if c(u0ur+1)=c(vr1vr).

    If r=3, then since v2u2 is an out-half-edge, c(v2v3)=c(u1u2)=c(u0u4), hence c can be extended to a good coloring of G.

    If r=4, then by Lemma 2.2, v3u3 is an out-half-edge, hence c(v3v4)=c(u2u3). Since c(u2u3)c(u1u2), it follows that c(u0ur+1)c(v3v4), hence c cannot be extended to a good coloring of G.

    If r=5, then c(v4v5)=c(u3u4) and c(u3u4)=¯c(v3) since v4u4 is still an out-half-edge. Note that ¯c(v3)c(v2v3) and c(v2v3)=c(u1u2), it follows that c(v4v5)c(u1u2), that is, c(u0ur+1)c(v4v5), hence c cannot be extended to a good coloring of G.

    If r6, then r24. By Lemma 2.3, let α=c(u0ur+1), then c can be extended to a good coloring of G2,r2 such that c(vr1vr)=c(u0ur+1), hence c can be extended to a good coloring of G.

    By symmetry, if the coloring of G1 is of type (4), then the edge v2u2 is an in-half-edge. By the same argument, we will obtain a good coloring of G if r4 and r5.

    In summary, if r{4,5}, we could obtained a good coloring of G, and for r=4 or r=5, χavd(G)5. Since G is cubic, χavd(G)4, thus χavd(G)=4 if r{4,5}. For r=4 or r=5, from Theorem 1.1, we have χavd(G)=5.

    Lemma 2.5. Suppose Gi,k is a bottom block, and Hi,k has a good coloring c such that viui is a full-edge. If k=1 or k3, then c can be extended to a good coloring of G. If k=2, then c can be extended to a good coloring of G if and only if c(ui1ur+1) is suitable.

    Proof. Without loss of generality, let c(vi1vi)=1,c(uiui)=2,¯c(vi1)=3 and ¯c(ui)=4. We will consider the following two cases.

    Case 1. k=1. Then i=r1. Let c(vr1vr)=c(ur1ur)=3, c(vr1ur1)=4. If c(ur2ur+1)=3, then statement (1) of Lemma 2.1 holds. If c(ur2ur+1)3, we have ¯c(ur2)c(ur1ur) since ¯c(ur2)¯c(vr2) and ¯c(vr2)=3. Hence statement (2) of Lemma 2.1 holds. Therefore, we will obtain a good coloring of G by Lemma 2.1.

    Case 2. k2. Note that ¯c(vi1) must appear on the edges incident with vi, that is, viui or vivi+1 is colored with 3.

    Subcase 2.1. viui is colored with 3. Then uiui+1 is colored with 4. If vivi+1 is colored with 4, then vi+1ui+1 is a crossing-edge, by Lemma 2.4, this coloring cannot be extended to a good coloring of G. Hence vivi+1 is colored with 2. It follows that ¯c(vi)=4 and ¯c(ui)=1. Thus vi+1ui+1 is an out-half-edge. By Lemma 2.2, c can be extended to a good coloring of G if and only if c(ui1ur+1)=c(vr1vr).

    If k=2, then r=i+2, c(vr1vr)=c(vi+1vi+2). Since c(vi+1vi+2)=c(uiui+1)=4, c can be extended to a good coloring of G if and only if c(ui1ur+1)=4=¯c(ui).

    If k=3, then r=i+3. Denote c(ui1ur+1)=α. If α{1,3}, then let c(vi+2vi+3)=c(ui+1ui+2)=α, c(vi+1vi+2)=4, c(vi+1ui+1)={1,3}{α}, c(ui+2vi+2){1,2,3}{α}, c(ui+2ur)={1,2,3}{α,c(ui+2vi+2)}. Now we obtain a good coloring of Gi,k such that c(ui1ur+1)=c(vi+2vi+3)=c(vr1vr).

    If k=4, then r=i+4. Denote c(ui1ur+1)=α. If α{1,2,3}, then let c(vi+1vi+2)=4, c(vi+3vi+4)=c(ui+2ui+3)=α, c(vi+2vi+3)=c(ui+1ui+2){1,3}{α}, c(ui+3ur){1,2,3,4}{c(vi+2vi+3),α}, c(vjuj)={1,2,3,4}{c(vj1vj),c(vjvj+1),c(ujuj+1)} for j=i+1,i+2,i+3. Now we obtain a good coloring of Gi,k such that c(ui1ur+1)=c(vi+3vi+4)=c(vr1vr).

    If k5, by Lemma 2.3, let α=c(ui1ur+1), then we can obtain a good coloring of Gi,k such that c(vr1vr)=α=c(ui1ur+1).

    Subcase 2.2. vivi+1 is colored with 3. If uiui+1 is colored with 4, then the edge viui cannot be colored to obtain a good coloring. Hence viui is colored with 4. If uiui+1 is colored with 3, then vi+1ui+1 is a crossing-edge, by Lemma 2.4, this coloring cannot be extended to a good coloring of G. Hence uiui+1 is colored with 1. It follows that ¯c(vi)=2 and ¯c(ui)=3. Thus vi+1ui+1 is an in-half-edge. By Lemma 2.2, c can be extended to a good coloring of G if and only if c(ui1ur+1)=c(ur1ur).

    If k=2, then r=i+2, c(ur1ur)=c(ui+1ui+2). Since c(ui+1ui+2)=c(vivi+1)=3, c can be extended to a good coloring of G if and only if c(ui1ur+1)=3=¯c(vi1).

    If k=3, then r=i+3. Denote c(ui1ur+1)=α. If α{2,4}, then let c(ui+2ui+3)=c(vi+1vi+2)=α, c(ui+1ui+2)=3, c(vi+1ui+1)={2,4}{α}, c(ui+2vi+2){1,2,4}{α},c(vi+2vr)={1,2,4}{α,c(ui+2vi+2)}. Now we obtain a good coloring of Gi,k such that c(ui1ur+1)=c(ui+2ui+3)=c(ur1ur).

    If k=4, then r=i+4. Denote c(ui1ur+1)=α. If α{1,2,4}, then let c(ui+1ui+2)=3, c(ui+3ui+4)=c(vi+2vi+3)=α, c(ui+2ui+3)=c(vi+1vi+2){2,4}{α}, c(vi+3vr){1,2,3,4}{c(ui+2ui+3),α}, c(vjuj)={1,2,3,4}{c(uj1uj),c(ujuj+1),c(vjvj+1)} for j=i+1,i+2,i+3. Now we obtain a good coloring of Gi,k such that c(ui1ur+1)=c(ui+3ui+4)=c(ur1ur).

    If k5, by Lemma 2.3, let α=c(ui1ur+1), then we can obtain a good coloring of Gi,k such that c(ur1ur)=α=c(ui1ur+1).

    Combining Subcase 2.1 and Subcase 2.2, for k3, we can obtain a good coloring of G. But for k=2, c can be extended to a good coloring of G if and only if c(ui1ur+1){¯c(ui),¯c(vi1)}, that is c(ui1ur+1) is suitable.

    Lemma 2.6. Let Gi,k,c be a k-crossing block. Suppose the associated subgraph Hi,k,c has a good coloring c such that viui is a full-edge.

    (1) If vi1ui1 is an outer-crossing-edge, then c can be extended to a good coloring of Gi,k,c such that for each j, i+1ji+k, vjuj is a full-edge and vj1uj1 is an outer-crossing-edge.

    (2) If vi1ui1 is a crossing-edge, then for each j, i+1ji+k, we can extend c such that vjuj is a full-edge and c(uj1uj1)=¯c(vj1).

    Proof. Without loss of generality, assume that c(vi1vi)=1,c(uiui)=2,¯c(vi1)=3 and ¯c(ui)=4.

    Considering the case that vi1ui1 is an outer-crossing-edge, that is, c(vi1vi)=¯c(ui1) and c(ui1ui+1)=¯c(ui). So ¯c(ui1)=1,c(ui1ui+1)=4. We set c(vivi+1)=3,c(uiui)=1, and c(viui)=4. Then ¯c(vi)=2 and ¯c(ui)=3. Hence, the edge vi+1ui+1 is a full-edge, and viui is an outer-crossing-edge. Note that the edge vi+1ui+1 has the same property as viui, then we can do the similar coloring such that for each j, i+1ji+k, vjuj is a full-edge and vj1uj1 is an outer-crossing-edge.

    Considering the case that vi1ui1 is a crossing-edge, that is, c(vi1vi)=¯c(ui1) and c(ui1ui+1)=¯c(vi1). So ¯c(ui1)=1,c(ui1ui+1)=3. If k=1, we set c(vivi+1)=2,c(uiui)=4, and c(viui)=3. Then ¯c(vi)=4 and ¯c(ui)=1. Hence, the edge vi+1ui+1 is a full-edge, and c(uiui)=¯c(vi). If k2, then we reset the coloring such that c(vi+1ui+1)=1, c(vi+1vi+2)=2, c(vivi+1)=c(uiui+2)=3, c(viui)=c(ui+1ui+1)=4. Then ¯c(vi+1)=4, ¯c(ui+1)=2, and ¯c(ui)=1. Hence, the edge vi+2ui+2 is a full-edge, and vi+1ui+1 is a crossing-edge, which shows that c(ui+1ui+1)=¯c(vi+1). Note that the edge vi+2ui+2 has the same property as viui, then we can do the similar coloring such that vjuj is a full-edge and c(uj1uj1)=¯c(vj1) for i+1ji+k.

    Lemma 2.7. Let Gi,k be a k-block with k2. Suppose Hi,k has a good coloring c such that viui is a full-edge, then

    (1) If Gi,k is adjacent to a t-block Gi+k,t, then we can extend the coloring c such that vi+kui+k is a full-edge. Moreover, if c(ui+k1ui+k1) is not suitable and the Gi+k,t is a bottom block with t=2, then c can be extended to a good coloring of G.

    (2) If Gi,k is adjacent to a t-crossing block Gi+k,t,c with t2, then we can extend the coloring c such that vi+k+t1ui+k+t1 is a full-edge and c(ui+k+t2ui+k+t2) is suitable.

    Proof. Without loss of generality, suppose c(vi1vi)=1,c(uiui)=2,¯c(vi1)=3 and ¯c(ui)=4. Then we have ¯c(ui1)3. If c(ui1ui1)=2 and ¯c(ui1)=4, then c(vi2vi1)=4 and c(ui1ui1)=3, it follows that the edge vi1ui1 cannot be AVD-edge-colored with 4 colors. Similarly for the case c(ui1ui1)=4 and ¯c(ui1)=2. Hence, we have c(ui1ui1),¯c(ui1){1,2,2,1,1,4,4,1,3,1,3,2,3,4}, where a,b is an ordered pair and a,b=c,d if and only if a=c and b=d.

    Now we divide the proof into the following three cases depending on k.

    Case 1. k=2. Then ui1=ui+2.

    Subcase 1.1. c(ui1ui+2),¯c(ui1){1,2,2,1}.

    First considering the case that Gi,2 is adjacent to a t-block. Set c(vi+1ui+1)=1,c(vivi+1)=2, c(viui)=c(ui+1ui+1)=3,c(uiui+1)=c(vi+1vi+2)=4, then ¯c(vi+1)=3 and c(ui+1ui+1)=¯c(vi+1). It is easy to see that vi+2ui+2 is a full-edge and c(ui+1ui+1) is suitable.

    Now considering the case that Gi,2 is adjacent to a t-crossing block. Set c(vi+1ui+1)=1,c(vi+1vi+2)=2, c(viui)=c(ui+1ui+1)=c(vi+2ui+2)=3,c(vivi+1)=c(uiui+1)=c(vi+2vi+3)=4. If c(ui1ui+2),¯c(ui1)=1,2, then set c(ui+2ui+2)=2. It is easy to see that vi+3ui+3 is a full-edge and vi+2ui+2 is an outer-crossing-edge. If c(ui1ui+2),¯c(ui1)=2,1, then set c(ui+2ui+2)=1. It follows that vi+3ui+3 is a full-edge and vi+2ui+2 is a crossing-edge. By Lemma 2.6, we can extend the coloring c such that vi+2+t1ui+2+t1 is a full-edge and c(ui+2+t2ui+2+t2) is suitable.

    Subcase 1.2. c(ui1ui+2),¯c(ui1){1,4,4,1}.

    In this case, set c(vi+1ui+1)=1,c(ui+1ui+1)=2, c(viui)=c(vi+1vi+2)=3,c(vivi+1)=c(uiui+1)=4. Then ¯c(vi+1)=2, vi+2ui+2 is a full-edge and vi+1ui+1 is a crossing-edge. Hence, if Gi,2 is adjacent to a t-block, then we have shown that vi+2ui+2 is a full-edge and c(ui+1ui+1) is suitable. If Gi,2 is adjacent to a t-crossing block, then by Lemma 2.6, we can extend the coloring c such that vi+2+t1ui+2+t1 is a full-edge and c(ui+2+t2ui+2+t2) is suitable.

    Subcase 1.3. c(ui1ui+2),¯c(ui1){3,1,3,2,3,4}.

    First set c(viui)=4,c(vivi+1)=c(uiui+1)=3. If c(ui1ui+2),¯c(ui1)=3,1, then set c(vi+1ui+1)=1,c(vi+1vi+2)=2,c(ui+1ui+1)=4. If c(ui1ui+2),¯c(ui1)=3,2, then set c(vi+1ui+1)=2,c(vi+1vi+2)=4,c(ui+1ui+1)=1. If c(ui1ui+2),¯c(ui1)=3,4, then set c(vi+1ui+1)=4,c(vi+1vi+2)=2,c(ui+1ui+1)=1. In all these cases, we have that vi+2ui+2 is a full-edge and vi+1ui+1 is a crossing-edge. Therefore, the conclusion holds for these subcases.

    Case 2. k=3. Then ui1=ui+3.

    Subcase 2.1. c(ui1ui+3),¯c(ui1){1,2,2,1,1,4,4,1}.

    First considering that Gi,3 is adjacent to a t-block. If c(ui1ui+3),¯c(ui1){1,2,2,1}, then set c(uiui+1)=c(vi+2ui+2)=1,c(vi+1vi+2)=c(ui+2ui+2)=2, c(vivi+1)=c(ui+1ui+2)=3,c(viui)=c(vi+1ui+1)=c(vi+2vi+3)=4, denote this coloring as (A). Under this coloring, we have c(vi+2vi+3),¯c(vi+2)=4,3, hence vi+3ui+3 is a full-edge. Note that if c(ui1ui+3),¯c(ui1)=1,2, then c(ui+2ui+2) is suitable, and vi+2ui+2 is an outer-crossing-edge. But if c(ui1ui+3),¯c(ui1)=2,1, then c(ui+2ui+2) is not suitable. For this case, if Gi+3,t is a bottom block with t=2, see Figure 5, then we color the edges of Gi,3Gi+3,2Gr as follows: c(uiui+1)=c(vi+2vi+3)=c(ui+3ui+4)=c(vi+5ui+6)=1, c(vi+1ui+1)=c(vi+2ui+2)=c(vi+4vi+5)=c(ui+5ui+6)=2, c(vivi+1)=c(ui+1ui+2)=c(vi+3ui+3)=c(vi+4ui+4)=c(vi+5ui+5)=3, c(viui)=c(vi+1vi+2)=c(ui+2ui+6)=c(vi+3vi+4)=c(ui+4ui+5)=4. It follows that c is a good coloring of G.

    Figure 5.  The three colorings of Gi,3.

    If c(ui1ui+3),¯c(ui1){1,4,4,1}, then set c(uiui+1)=c(vi+2ui+2)=1,c(vi+1ui+1)=c(vi+2vi+3)=2,, c(vivi+1)=c(ui+1ui+2)=3,c(viui)=c(vi+1vi+2)=c(ui+2ui+2)=4, denote this coloring as (B). Under this coloring, we have c(vi+2vi+3),¯c(vi+2)=2,3, hence vi+3ui+3 is a full-edge. Similarly, if c(ui1ui+3),¯c(ui1)=1,4, then c(ui+2ui+2) is suitable, and vi+2ui+2 is an outer-crossing-edge. But if c(ui1ui+3),¯c(ui1)=4,1, then c(ui+2ui+2) is not suitable. For this case, if Gi+3,t is a bottom block with t=2, see Figure 6, then we color the edges of Gi,3Gi+3,2Gr as follows: c(uiui+1)=c(vi+2vi+3)=c(ui+3ui+4)=c(vi+5ui+6)=1, c(vi+1vi+2)=c(ui+2ui+6)=c(vi+3vi+4)=c(ui+4ui+5)=2, c(vivi+1)=c(ui+1ui+2)=c(vi+3ui+3)=c(vi+4ui+4)=c(vi+5ui+5)=3, c(viui)=c(vi+1ui+1)=c(vi+2ui+2)=c(vi+4vi+5)=c(ui+5ui+6)=4. It follows that c is a good coloring of G.

    Figure 6.  The 3-block adjacent with a 2-bottom block.

    Now considering the case that Gi,3 is adjacent to a t-crossing block. If c(ui1ui+3),¯c(ui1)=1,2, then we use coloring (A), under this coloring, vi+3ui+3 is a full-edge and vi+2ui+2 is an outer-crossing-edge, by Lemma 2.6, we can extend the coloring c such that vi+3+t1ui+3+t1 is a full-edge and c(ui+3+t2ui+3+t2) is suitable. Similarly, if c(ui1ui+3),¯c(ui1)=1,4, then we use coloring (B), and extend this coloring to the t-crossing block such that vi+3+t1ui+3+t1 is a full-edge and c(ui+3+t2ui+3+t2) is suitable.

    For the case c(ui1ui+3),¯c(ui1)=2,1, we first use coloring (B), and then set c(vi+3ui+3)=4,c(ui+3ui+3)=1, and c(vi+3vi+4)=3, then ¯c(vi+3)=1,¯c(ui+3)=3, so vi+3ui+3 is a crossing-edge. Note that vi+4ui+4 is a full-edge. By Lemma 2.6, we can extend the coloring c such that vi+3+t1ui+3+t1 is a full-edge and c(ui+3+t2ui+3+t2) is suitable. Similarly, for the case c(ui1ui+3),¯c(ui1)=4,1, we first use coloring (A), and then set c(vi+3ui+3)=2,c(ui+3ui+3)=1, and c(vi+3vi+4)=3, which makes vi+3ui+3 a crossing-edge and vi+4ui+4 a full-edge. By Lemma 2.6, the conclusion holds for this case.

    Subcase 2.2. c(ui1ui+3),¯c(ui1)=3,4.

    Considering that Gi,3 is adjacent to a t-block. Set c(ui+1ui+2)=c(vi+2vi+3)=1,c(vivi+1)=c(ui+2ui+2)=2, c(viui)=c(vi+1ui+1)=c(vi+2ui+2)=3,c(uiui+1)=c(vi+1vi+2)=4, denote this coloring as (C). Then c(vi+2vi+3),¯c(vi+2)=1,2, hence vi+3ui+3 is a full-edge and c(ui+2ui+2) is suitable.

    Considering the case that Gi,k is adjacent to a t-crossing block. If t=2, then let c(uiui+1)=c(vi+2vi+3)=1,c(vi+1ui+1)=c(vi+2ui+2)=c(ui+3ui+3)=2, c(vivi+1)=c(ui+1ui+2)=c(vi+3vi+4)=3,c(viui)=c(vi+1vi+2)=c(vi+3ui+3)=c(ui+2ui+2)=4. Then vi+4ui+4 becomes a full-edge and c(ui+3ui+3) is suitable. If t3, then we first use coloring (C), then set c(vi+4ui+4)=c(ui+3ui+5)=1,c(vi+3vi+4)=2, c(vi+4vi+5)=3,c(vi+3ui+3)=c(ui+4ui+4)=4. It is easy to see that vi+5ui+5 is a full-edge and vi+4ui+4 is a crossing-edge. By Lemma 2.6, we can extend the coloring c such that vi+3+tui+3+t is a full-edge and c(ui+3+t1ui+3+t1) is suitable.

    Subcase 2.3. c(ui1ui+3),¯c(ui1)=3,2.

    Since ¯c(ui1)=2 and ¯c(vi1)=3, we have c(vi2vi1)=2 and c(vi1ui1)=4. If ¯c(vi2)4, then we transform c(ui1ui+3) from 3 to 4 and c(ui1vi1) from 4 to 3, which changes ¯c(vi1) from 3 to 4. Set c(uiui+1)=c(vi+2vi+3)=1,c(vi+1vi+2)=c(ui+2ui+2)=2, c(vivi+1)=c(ui+1ui+2)=3,c(viui)=c(vi+1ui+1)=c(vi+2ui+2)=4. Then vi+3ui+3 becomes a full-edge and vi+2ui+2 becomes an outer-crossing-edge. If ¯c(vi1)=4, then we transform c(vi1vi) from 1 to 3, which changes ¯c(vi) from 3 to 1. Note that viui is still a full-edge, we exchange the color 1 and 3 in coloring (A), it follows that vi+3ui+3 becomes a full-edge and vi+2ui+2 becomes an outer-crossing-edge. Hence the conclusion holds for this subcase.

    Subcase 2.4. c(ui1ui+3),¯c(ui1)=3,1.

    In this case, c(vi1ui1){2,4}.

    Subcase 2.4.1. c(vi1ui1)=4. Then c(ui1ui1)=c(vi2vi1)=2.

    We only need to consider the case that at least one of ¯c(ui1) and ¯c(vi2) is distinct with 4. Otherwise, if ¯c(ui1)=¯c(vi2)=4, then vi2ui1E(G), hence vi2uiE(G), but ¯c(ui)=4, which is impossible.

    Subcase 2.4.1.1. ¯c(ui1)4.

    Consider the coloring c on Hi,3, we transform c(vi1ui1) from 4 to 1 and c(vi1vi) from 1 to 4, then c(ui1ui+3),¯c(ui1)=3,4, c(vi1vi),¯c(vi1)=4,3. If Gi,3 is adjacent to a t-block, then set c(vivi+1)=c(ui+1ui+2)=1,c(vi+1ui+1)=c(vi+2vi+3)=2, c(viui)=c(vi+1vi+2)=c(ui+2ui+2)=3,c(uiui+1)=c(vi+2ui+2)=4. Then vi+3ui+3 becomes a full-edge, but c(ui+2ui+2) is not suitable. For this case, if Gi+3,t is a bottom block with t=2, then we color the edges of Gi,3Gi+3,2Gr as follows: c(viui)=c(vi+1ui+1)=c(vi+2vi+3)=c(ui+4ui+5)=c(vi+5ui+6)=1, c(vi+1vi+2)=c(ui+2ui+6)=c(ui+3ui+4)=c(vi+4vi+5)=2, c(vivi+1)=c(ui+1ui+2)=c(vi+3vi+4)=c(ui+5ui+6)=3, c(uiui+1)=c(vi+2ui+2)=c(vi+3ui+3)=c(vi+4ui+4)=c(vi+5ui+5)=4. Then c is a good coloring of G.

    If Gi,3 is adjacent to a t-crossing block, then set c(vivi+1)=c(ui+1ui+2)=c(vi+3vi+4)=1,c(vi+1vi+2)=c(ui+2ui+4)=c(vi+3ui+3)=2, c(viui)=c(vi+1ui+1)=c(vi+2vi+3)=3,c(vi+2ui+2)=c(ui+3ui+3)=4. Then vi+4ui+4 becomes a full-edge and vi+3ui+3 becomes a crossing-edge. By Lemma 2.6, we can extending the coloring c such that vi+3+t1ui+3+t1 is a full-edge and c(ui+3+t2ui+3+t2) is suitable.

    Subcase 2.4.1.2. ¯c(vi2)4.

    Consider the coloring c on Hi,3, we transform c(ui1ui+3) from 3 to 4 and c(vi1ui1) from 4 to 3, then c(ui1ui+3),¯c(ui1)=4,1, c(vi1vi),¯c(vi1)=1,4. If Gi,3 is adjacent to a t-block, then we use coloring (B) on Gi,k, under this coloring, vi+3ui+3 is a full-edge, but c(ui+2ui+2) is not suitable. For this case, if Gi+3,t is a bottom block with t=2, then we give a coloring on the edges of Gi,3Gi+3,2Gr as follows: c(uiui+1)=c(vi+2vi+3)=c(ui+3ui+4)=c(vi+5ui+6)=1, c(vi+1vi+2)=c(ui+2ui+6)=c(vi+3vi+4)=c(ui+4ui+5)=2, c(vivi+1)=c(ui+1ui+2)=c(vi+3ui+3)=c(vi+4vi+5)=c(ui+5ui+6)=3, c(viui)=c(vi+1ui+1)=c(vi+2ui+2)=c(vi+4ui+4)=c(vi+5ui+5)=4. Then c is a good coloring of G.

    If Gi,3 is adjacent to a t-crossing block with t2, then we use coloring (A) on Gi,3, and set c(ui+3ui+3)=1,c(vi+3ui+3)=2, c(vi+3vi+4)=3. It is easy to see that vi+4ui+4 is a full-edge and vi+3ui+3 becomes a crossing-edge. Hence the conclusion holds for this subcase.

    Subcase 2.4.2. c(vi1ui1)=2. Then c(ui1ui1)=c(vi2vi1)=4.

    We divide the proof of this case into the following two parts depending on which side ui2 is on.

    Subcase 2.4.2.1. ui2 and ui1 are on the same side of P, that is, ui1=ui2.

    Since c(ui1ui1)=c(vi2vi1)=4, we have c(vi2ui2){1,2,3}.

    If c(vi2ui2)=1, then under the coloring c on Hi,3, we transform c(vi2ui2),c(vi1vi) from 1 to 4, and c(ui2ui1),c(vi2vi1) from 4 to 1. Note that c is still a good coloring of Hi,k, and c(ui1ui+3),¯c(ui1)=3,4, c(vi1vi),¯c(vi1)=4,3. We then use the same coloring with subcase 2.4.1.1.

    If c(vi2ui2)=2, then consider the coloring c on Hi,k, we transform c(vi2ui2),c(vi1ui1) from 2 to 4, and c(ui2ui1),c(vj2vj1) from 4 to 2, then c is still a good coloring of Hi,k, and it is the subcase 2.4.1.

    If c(vi2ui2)=3, then under the coloring c on Hi,k, we transform c(vi2ui2),c(ui1ui1) from 3 to 4, and c(ui2ui1),c(vi2vi1) from 4 to 3, then c(ui1ui+3),¯c(ui1)=4,1, c(vi1vi),¯c(vi1)=1,4. We then use the same coloring with subcase 2.4.1.2.

    Subcase 2.4.2.2. ui2 and ui are on the same side of P, that is, ui=ui2.

    Since ¯c(ui2)=4 and c(ui2ui)=2, we have c(vi2ui2){1,3}.

    Considering the case that c(vi2ui2)=1, then c(ui2ui2)=3. We also have c(vi3vi2)=3 since ¯c(vi1)=3. We may assume that ¯c(ui2)=1. Otherwise if ¯c(ui2)1, then we transform c(vi1vi),c(vi2ui2) from 1 to 4, and c(vi2vi1) from 4 to 1, which changes ¯c(ui2) from 4 to 1. We turn to subcase 2.2, and exchange the color 1 and 4 in the coloring of Gi,k or Gi,kGi+k,t,c, then we obtain the desired coloring. Consider ¯c(vi3), it cannot equal to c(vi3vi2) and ¯c(vi2), hence ¯c(vi3){1,4}.

    If ¯c(vi3)=1, then ui3=ui1. Since ¯c(ui1)=¯c(vi3)=1 and c(vi3vi2)=3, we have c(ui3ui3)=1, c(vi3ui3)=2 and c(vi4vi3)=4. If ¯c(ui3)4, then we transform c(ui3ui1) from 4 to 3, and transform c(ui1ui+3) from 3 to 4, which can turn to subcase 2.1 for the case c(ui1ui+3),¯c(ui1)=4,1. If ¯c(ui3)=4, then we transform c(vi3ui3),c(vi1ui1) from 2 to 3, c(vi3vi2),c(ui1ui+3) from 3 to 2, and transform c(ui2ui) from 4 to 2, then c(ui1ui+3),¯c(ui1)=2,1, c(vi1vi),¯c(vi1)=1,2. We exchage the color 2 and 3 in the subcase 2.4.1 for the case ¯c(vi2)4.

    If ¯c(vi3)=4, consider ¯c(ui1), it cannot equal to c(ui1ui1) and ¯c(ui1), hence ¯c(ui1){2,3}. If ¯c(ui1)=2, then we transform c(vi1ui1),c(ui2ui) from 2 to 1, c(vi1vi) from 1 to 3, c(ui1ui+3) from 3 to 2, c(vi2vi1) from 4 to 2, c(vi2ui2) from 1 to 4, then we have ¯c(ui1)=3, ¯c(vi1)=4 and ¯c(ui2)=2. We turn to subcase 2.1 when c(ui1ui+3),¯c(ui1)=4,1, and replace the color 1 to 3, 3 to 4, 4 to 2, and 2 to 1 in the coloring of Gi,k or Gi,kGi+k,t,c, then we obtain the desired coloring. If ¯c(ui1)=3, then we transform c(vi1ui1),c(ui2ui) from 2 to 1, c(vi2ui2),c(vi1vi) from 1 to 4, c(vi2vi1) from 4 to 2. We turn to subcase 2.2, replace the color 1 to 4, 4 to 2, and 2 to 1 in the coloring of Gi,k or Gi,kGi+k,t,c, then we obtain the desired coloring.

    Now we consider the case that c(vi2ui2)=3. Since ¯c(ui2)=4, we have c(ui2ui2)=1. Note that c(vi3vi2){1,2}. For the case c(vi3vi2)=1, consider ¯c(ui2), we may assume ¯c(ui2)=3. Otherwise, if ¯c(ui2)3, then we transform c(vi2ui2) from 3 to 4, and transform c(vi2vi1) from 4 to 3, it follows that ¯c(vi1)=4 and ¯c(ui2)=3. We turn to subcase 2.1 when c(ui1ui+3),¯c(ui1)=4,1, and exchange the color 3 and 4 in the coloring of Gi,k or Gi,kGi+k,t,c, then we obtain the desired coloring. Now consider ¯c(vi3), it can be 3 or 4. If ¯c(vi3)=3, then ui3=ui1, and c(ui3ui3)=3. But since ¯c(ui1)=1 and c(vi3vi2)=1, it implies that c(ui3ui3)=1, a contradiction. Hence ¯c(vi3)3, then ¯c(vi3)=4. In this case, we transform c(vi1ui1),c(ui2ui) from 2 to 3, c(vi2ui2) from 3 to 4, c(ui1ui+3) from 3 to 2, and c(vi2vi1) from 4 to 2. We turn to subcase 2.1 when c(ui1ui+3),¯c(ui1)=4,1, and replace the color 2 to 3, 3 to 4, 4 to 2 in the coloring of Gi,k or Gi,kGi+k,t,c, then we obtain the desired coloring.

    For the case c(vi3vi2)=2, consider ¯c(ui2), it can be 2 or 3. If ¯c(ui2)=2, then we transform c(vi2vi1) from 4 to 3, and transform c(vi2ui2) from 3 to 4, then turn to subcase 2.1 when c(ui1ui+3),¯c(ui1)=4,1, and exchange the color 3 and 4 in the coloring of Gi,k or Gi,kGi+k,t,c, then we obtain the desired coloring. If ¯c(ui2)=3, then we transform c(ui2ui) from 2 to 4, c(vi1ui1) from 2 to 3, c(ui1ui+3) from 3 to 2, and turn to subcase 2.4.1.2, exchange the color 2 and 4 in the coloring of Gi,k or Gi,kGi+k,t,c, then we obtain the desired coloring.

    Case 3. k4.

    Let c(ui1ui1),¯c(ui1)=α,β, and α,β be an ordered pair in {1,2,2,1,1,4,4,1, 3,1,3,2,3,4}.

    Subcase 3.1. k=4.

    Let γ be a color in {2,4}{α,β}, δ={1,2,3,4}{α,β,γ}. Set c(viui)=4,c(uiui+1)=1,c(vivi+1)=c(ui+1ui+2)=3, c(vi+1vi+2)=c(ui+2ui+3)=γ,c(vi+2vi+3)=c(ui+3ui+3)=β, c(vi+3vi+4)=δ. For i+1ji+3, let c(vjuj)={1,2,3,4}{c(vj1vj),c(vjvj+1),c(uj1uj)}. Then, vi+4ui+4 is a full-edge and vi+3ui+3 is an outer-crossing-edge. Hence, if Gi,k is adjacent to a t-block, then vi+kui+k is a full-edge and c(ui+k1ui+k1) is suitable. If Gi,k is adjacent to a t-crossing block with t2, by Lemma 2.6, we can extend the coloring c such that vi+k+t1ui+k+t1 is a full-edge and c(ui+k+t2ui+k+t2) is suitable.

    Subcase 3.2. k=5.

    If α1 and β1, then set c(viui)=4,c(uiui+1)=1,c(vivi+1)=c(ui+1ui+2)=3, c(vi+2vi+3)=c(ui+3ui+4)=1,c(vi+3vi+4)=c(ui+4ui+4)=β, c(vi+1vi+2)=c(ui+2ui+3){2,4}{β}, c(vi+4vi+5){1,2,3,4}{1,α,β}. For i+1ji+4, let c(vjuj)={1,2,3,4}{c(vj1vj),c(vjvj+1),c(uj1uj)}. Then, vi+5ui+5 is a full-edge and vi+3ui+3 is an outer-crossing-edge.

    If α1 and β=1, then set c(viui)=4,c(uiui+1)=1,c(vivi+1)=c(ui+1ui+2)=3, c(vi+3vi+4)=c(ui+4ui+4)=1, c(vi+2vi+3)=c(ui+3ui+4){2,4}{α}, c(vi+1vi+2)=c(ui+2ui+3){1,2,3,4}{1,3,c(vi+2vi+3)}, c(vi+4vi+5){1,2,3,4}{1,α,c(vi+2vi+3)}. For i+1ji+4, let c(vjuj)={1,2,3,4}{c(vj1vj),c(vjvj+1),c(uj1uj)}. Then, vi+5ui+5 is a full-edge and vi+3ui+3 is an outer-crossing-edge.

    If α=1, then β may be 2 or 4. Consider α,β=1,2. If Gi,k is adjacent to a t-block with t1, then set c(ui+1ui+2)=c(vi+2vi+3)=c(vi+4ui+4)=1, c(vivi+1)=c(vi+3vi+4)=c(ui+2ui+3)=2, c(viui)=c(vi+1ui+1)=c(vi+2ui+2)=c(ui+3ui+4)=c(vi+4vi+5)=3, c(uiui+1)=c(vi+1vi+2)=c(vi+3ui+3)=c(ui+4ui+4)=4, hence vi+5ui+5 is a full-edge and c(ui+4ui+4) is suitable. If Gi,k is adjacent to a t-crossing block with t2, then set c(uiui+1)=c(vi+2ui+2)=c(vi+3ui+3)=c(vi+4vi+5)=1, c(vi+1vi+2)=c(ui+2ui+3)=c(vi+4ui+4)=c(ui+5ui+5)=2, c(vivi+1)=c(ui+1ui+2)=c(vi+3vi+4)=c(ui+4ui+6)=c(vi+5ui+5)=3, c(viui)=c(vi+1ui+1)=c(vi+2vi+3)=c(ui+3ui+4)=c(vi+5vi+6)=4, hence vi+6ui+6 is a full-edge and vi+5ui+5 is a crossing-edge, by Lemma 2.6, the conclusion holds for this case.

    Consider α,β=1,4. If Gi,k is adjacent to a t-block, then set c(vi+1ui+1)=c(ui+2ui+3)=c(vi+3vi+4)=1, c(vivi+1)=c(vi+2ui+2)=c(ui+3ui+4)=c(vi+4vi+5)=2, c(viui)=c(ui+1ui+2)=c(vi+2vi+3)=c(ui+4ui+4)=3, c(uiui+1)=c(vi+1vi+2)=c(vi+3ui+3)=c(vi+4ui+4)=4, hence vi+5ui+5 is a full-edge and c(ui+4ui+4) is suitable. If Gi,k is adjacent to a t-crossing block with t2, then set c(uiui+1)=c(vi+2ui+2)=c(vi+3ui+3)=c(vi+4vi+5)=1, c(vi+1ui+1)=c(vi+2vi+3)=c(ui+3ui+4)=c(vi+5vi+6)=2, c(vivi+1)=c(ui+1ui+2)=c(vi+3vi+4)=c(ui+4ui+6)=c(vi+5ui+5)=3, c(viui)=c(vi+1vi+2)=c(vi+4ui+4)=c(ui+5ui+5)=4, hence vi+6ui+6 is a full-edge and vi+5ui+5 is a crossing-edge, by Lemma 2.6, the conclusion holds for this case.

    Subcase 3.3. k=6.

    First assume that {α,β}{1,3}, set c(viui)=4,c(uiui+1)=1,c(vivi+1)=c(ui+1ui+2)=3, c(vi+4vi+5)=c(ui+5ui+5)=β, c(vi+3vi+4)=c(ui+4ui+5){1,3}{α,β}, c(vi+2vi+3)=c(ui+3ui+4){1,2,3,4}{3,β,c(vi+3vi+4)}, c(vi+1vi+2)=c(ui+2ui+3){1,2,3,4}{1,3,c(vi+2vi+3)}, c(vi+5vi+6){1,2,3,4}{α,β,c(vi+3vi+4)}. For i+1ji+5, let c(vjuj)={1,2,3,4}{c(vj1vj),c(vjvj+1),c(uj1uj)}. Then, vi+5ui+5 is a full-edge and vi+3ui+3 is an outer-crossing-edge.

    Consider that {α,β}={1,3}, since ¯c(ui1)3, we have α=3, β=1. If Gi,k is adjacent to a t-block, then set c(ui+1ui+2)=c(vi+2vi+3)=c(vi+4ui+4)=c(vi+5ui+5)=1, c(vivi+1)=c(ui+2ui+3)=c(vi+3vi+4)=c(ui+5ui+5)=2, c(viui)=c(vi+1ui+1)=c(vi+2ui+2)=c(ui+3ui+4)=c(vi+4vi+5)=3, c(uiui+1)=c(vi+1vi+2)=c(vi+3ui+3)=c(ui+4ui+5)=c(vi+5vi+6)=4, hence vi+6ui+6 is a full-edge and c(ui+5ui+5) is suitable. If Gi,k is adjacent to a t-crossing block with t2, then set c(uiui+1)=c(vi+2vi+3)=c(ui+3ui+4)=c(vi+5ui+5)=c(ui+6ui+6)=1, c(vi+1vi+2)=c(ui+2ui+3)=c(vi+4vi+5)=c(ui+5ui+7)=c(vi+6ui+6)=2, c(vivi+1)=c(ui+1ui+2)=c(vi+3ui+3)=c(vi+4ui+4)=c(vi+5vi+6)=3, c(viui)=c(vi+1ui+1)=c(vi+2ui+2)=c(vi+3vi+4)=c(ui+4ui+5)=c(vi+6vi+7)=4, hence vi+7ui+7 is a full-edge and vi+6ui+6 is a crossing-edge, by Lemma 2.6, the conclusion holds for this case.

    Subcase 3.4. k7.

    We first set c(viui)=4,c(vivi+1)=3,c(uiui+1)=1, c(vi+k2vi+k1)=β. Since α,β{2,4,4,2}, we may set c(vi+k3vi+k2){2,4}{α,β}. For j=i+k4,i+k5,,i+3, let c(vjvj+1){2,3,4}{c(vj+1vj+2)),c(vj+2vj+3)}, then let c(vi+2vi+3)=1, c(vi+1vi+2){1,2,3,4}{1,3,c(vi+3vi+4)}. For i+1ji+k2, let c(ujuj+1)=c(vj1vj), c(vjuj)={1,2,3,4}{c(vj1vj),c(vjvj+1),c(uj1uj)}. Finally, let c(ui+k1ui+k1)=β, c(vi+k1vi+k)={1,2,3,4}{α,β,(vi+k3vi+k2)}, and c(vi+k1ui+k1)={1,2,3,4}{c(vi+k2vi+k1),c(vi+k1vi+k),c(ui+k2ui+k1)}. It is easy to see that c is a good coloring, and ¯c(vj)=c(uj1uj),c(vjvj+1)=¯c(uj) for i+1ji+k1. That is, we have c(vi+k1vi+k)=¯c(ui+k1). Hence vi+k1ui+k1 is an outer-crossing-edge since c(ui+k1ui+k1)=β. Note that, ¯c(vi+k1)=c(ui+k2ui+k1)=c(vi+k3vi+k2), hence ¯c(vi+k1){α,β}. By the fact that c(vi+k1vi+k){α,β} and c(vi+k1vi+k)¯c(vi+k1), we have vi+kui+k is a full-edge. If Gi,k is adjacent to a t-block, then vi+kui+k is a full-edge and c(ui+k1ui+k1) is suitable. If Gi,k is adjacent to a t-crossing block with t2, by Lemma 2.6, we can extend the coloring c such that vi+k+t1ui+k+t1 is a full-edge and c(ui+k+t2ui+k+t2) is suitable.

    Remark 2.2. By Lemma 2.6 and the proof of Lemma 2.7, if Gi,k is adjacent to a t-crossing block Gi+k,t,c with t2 and vi+k+tvr, then we can extend the coloring c such that vi+k+tui+k+t is a full-edge and c(ui+k+t1ui+k+t1) is suitable. Moreover, if Gi,k is adjacent to 1-block, then we can extend the coloring c such that vi+k+1ui+k+1 is a full-edge and c(ui+kui+k) is suitable.

    Lemma 2.8. Suppose Gi,k is not a bottom block and k2. If Hi,k has a good coloring c such that viui is a full-edge, then c can be extended to a good coloring of G.

    Proof. Assume that Gi,k is adjacent to Gi+k,t. By Lemma 2.7, we can extend c to a good coloring of Gi,k such that vi+kui+k is a full-edge. If Gi+k,t is a bottom block, then by Lemma 2.5, c can be extended to a good coloring of G if t=1, or t3, or t=2 and c(ui+k1ui+k1) is suitable. For t=2 and c(ui+k1ui+k1) is not suitable, by Lemma 2.7, c can also be extended to a good coloring of G.

    Therefore, we assume that Gi+k,t is not a bottom block. If t2, then the argument is similar as above. So assume that t=1, that is, vi+kui+k is in a crossing block. Suppose vi+kui+k is in a l-crossing block and l is maximal, that is, vi+k+l1ui+k+l1=vr1ur1 or vi+k+lui+k+l is in a d-block with d2. For l=1, since Gi+k,t is not a bottom block, vi+k+1ui+k+1 is in a d-block with d2. By Remark 2.2, we can extend the coloring c such that vi+k+1ui+k+1 is a full-edge and c(ui+kui+k) is suitable. If this d-block is a bottom block, then we can extend c to a good coloring of G by Lemma 2.5. If this d-block is not a bottom block, then we make the same argument as the case Gi,k. If l2, then by Lemma 2.7, we can extend c to a good coloring of Gi+k,l,c such that vi+k+l1ui+k+l1 is a full-edge and c(ui+k+l2ui+k+l2) is suitable. Hence, if vi+k+l1ui+k+l1=vr1ur1, then by Lemma 2.5, c can be extended to a good coloring of G. For the case vi+k+lui+k+l is in a d-block with d2, by Remark 2.2, we can extend c such that vi+k+tui+k+t is a full-edge and c(ui+k+t1ui+k+t1) is suitable. Hence if this d-block is a bottom block, then we can extend c to a good coloring of G by Lemma 2.5. If this d-block is not a bottom block, then we make the same argument as the case Gi,k.

    Corollary 2.1. Let Gi,k be a k-block with k2. If Hi,k has a good coloring c such that viui is a full-edge and c(ui1ui1) is suitable, then c can be extended to a good coloring of G.

    Corollary 2.2. Let Gi,k be a k-block. If Hi,k has a good coloring c such that viui is a full-edge and vi1ui1 is a crossing-edge or an outer-crossing-edge, then c can be extended to a good coloring of G.

    Proof. Note that c(ui1ui1) is suitable since vi1ui1 is a crossing-edge or an outer-crossing-edge. So if Gi,k is a bottom block, then c can be extended to a good coloring of G by Lemma 2.5. If Gi,k is not a bottom block and k2, by Lemma 2.8, we can extend c to a good coloring of G. Hence we only need to consider that Gi,k is not a bottom block and k=1. For this case, viui is in a crossing block, with the similar argument as Lemma 2.8, we can extend c + to a good coloring of G.

    Theorem 2.2. For every cubic Halin graph G in Gr which is not a necklace Nr, χavd(G)=4.

    Proof. Since G is not a necklace Nr, there are at least two blocks in G. Suppose the block containing v2u2 is a k-block.

    Case 1. k=1.

    In this case, u2 and u3 are on the different sides of P. We set c(v1u1)=c(v2u2)=1,c(v1u0)=c(u2u2)=2,c(u0u1)=c(v2v3)=3,c(u1u2)=c(v1v2)=c(u0u3)=4. Then v3u3 becomes a full-edge and v2u2 becomes a crossing-edge. By Corollary 2.2, c can be extended to a good coloring of G.

    Case 2. k=2.

    Suppose the block G2,k is adjacent to G4,t. If t2, then let c(v1u1)=c(v2u2)=c(v3u3)=1,c(v1u0)=c(u2u3)=c(v3v4)=2, c(u0u1)=c(v1v2)=c(u3u3)=3,c(u1u2)=c(v2v3)=c(u0u4)=4. Then v4u4 becomes a full-edge and c(u3u3) is suitable. Hence, by Corollary 2.1, we can extend c to a good coloring of G.

    For t=1, that is u5 and u4 are on the different sides of P. If v5u5 is in a l-block with l2, then let c(u1v1)=c(v2v3)=c(u3u5)=c(v4u4)=1, c(v1u0)=c(u1u2)=c(v3v4)=2, c(u0u1)=c(v2u2)=c(v3u3)=c(u4u4)=3, c(v1v2)=c(u0u4)=c(u2u3)=c(v4v5)=4. Then v5u5 becomes a full-edge and c(v4u4) is suitable. Hence, by Corollary 2.1, we can extend c to a good coloring of G. Therefore, v5u5 is in a 1-block, which means u6 and u5 are on the different sides of P.

    Consider the edge v6u6 is in a d-block, let c(u1v1)=c(u2u3)=c(v3v4)=c(u4u6)=c(v5u5)=1, c(v1u0)=c(v2u2)=c(v3u3)=c(v4v5)=2, c(u0u1)=c(v1v2)=c(u3u5)=c(v4u4)=3, c(u1u2)=c(u0u4)=c(v2v3)=c(v5v6)=c(u5u5)=4. Then v6u6 becomes a full-edge while c(u4u5) is not suitable. If G6,d is not a bottom block and d2, then by Lemma 2.8, we can extend c to a good coloring of G. If G6,d is a bottom block with d=1 or d3, then by Lemma 2.5, we can extend c to a good coloring of G. If G6,d is a bottom block with d=2, then G is the graph H0 depited in Figure. For this case, we give a good coloring of H0 as follows: let c(v1u1)=c(v2u2)=c(u3u5)=c(v4v5)=c(u4u6)=c(v7v8)=c(u8u9)=1, c(v1u0)=c(u2u3)=c(v3v4)=c(v5v6)=c(u6u7)=c(v8u9)=2, c(u1u0)=c(v1v2)=c(v3u3)=c(v4u4)=c(u5u9)=c(v6v7)=c(u7u8)=3, c(u1u2)=c(u0u4)=c(v2v3)=c(v5u5)=c(v6u6)=c(v7u7)=c(v8u8)=4.

    Now we consider that G6,d is not a bottom block and d=1. Let c(u1v1)=c(u2u3)=c(v3v4)=c(u4u6)=c(u5u7)=c(v5v6)=1, c(v1u0)=c(v2u2)=c(v3u3)=c(v4v5)=c(v6u6)=2, c(u0u1)=c(v1v2)=c(u3u5)=c(v4u4)=c(v6v7)=3, c(u1u2)=c(u0u4)=c(v2v3)=c(v5u5)=c(u6u6)=4. Then v7u7 becomes a full-edge and v6u6 becomes a crossing-edge. By Corollary 2.2, we can extend c to a good coloring of G.

    Case 3. k3.

    We first give a good coloring of G1 as follows: c(u1v1)=1,c(v1u0)=2,c(u0u1)=3, c(u1u2)=2,c(v1v2)=c(u0u2+k)=4. Then c(u0u2+k),¯c(u0)=4,1. Now we color the block G2,k.

    For k=3, set c(v2u2)=c(v3v4)=c(u4u4)=1, c(v3u3)=c(v4v5)=2, c(v2v3)=c(u3u4)=3, and c(u2u3)=c(v4u4)=4, then v5u5 is a full-edge and v4u4 is an outer-crossing edge.

    For k=4, set c(v2u2)=c(v3u3)=c(v4v5)=c(u5u5)=1, c(v3v4)=c(u4u5)=2, c(v2v3)=c(u3u4)=c(v5v6)=3, and c(u2u3)=c(v4u4)=c(v5u5)=4, then v6u6 is a full-edge and v5u5 is an outer-crossing edge.

    For k=5, set c(v2v3)=c(u3u4)=c(v5v6)=c(u6u6)=1, c(v3u3)=c(v4v5)=c(u5u6)=2, c(v2u2)=c(v3v4)=c(u4u5)=c(v6v7)=3, and c(u2u3)=c(v4u4)=c(v5u5)=c(v6u6)=4, then v7u7 is a full-edge and v6u6 is an outer-crossing edge.

    For k6, first set c(v2v3)=1, c(v3v4)=3. For 4jk2, let c(vjvj+1){1,2,3}{c(vj2vj1),c(vj1vj)}. Then let c(vk2vk1)=4, c(vk1vk){2,3}{c(vk3vk2)}, c(vkvk+1)=1, c(vk+1vk+2){2,3}{c(vk1vk)}. For 2jk, let c(ujuj+1)=c(vj1vj) and c(vjuj){1,2,3,4}{c(vj1vj),c(vjvj+1),c(uj1uj)}. Finally, let c(uk+1uk+1)=1 and c(vk+1uk+1)=4. Then c is a good coloring of G2,k, vk+2uk+2 is a full-edge and vk+1uk+1 is an outer-crossing edge.

    By Corollary 2.2, we can extend c to a good coloring of G.

    In summary, we have χavd(G)4. On the other hand, since G is cubic, χavd(G)4. Therefore, χavd(G)=4.

    Combining Theorem 2.1 and Theorem 2.2, we complete the proof of Theorem 1.2.

    In this paper, we have determined the exact values of the adjacent vertex distinguishing (AVD) chromatic indices of cubic Halin graphs whose characteristic trees are caterpillars. We showed that only two graphs have AVD chromatic index 5. For the cubic Halin graphs whose characteristic trees are not caterpillars, we believe that there are few graphs obtaining AVD chromatic index 5. It is interesting to figure out which cubic Halin graphs with characteristic trees not caterpillars have AVD chromatic index 5.

    The authors declare they have not used Artificial Intelligence (AI) tools in the creation of this article.

    The authors are grateful to reviewers for their valuable comments and suggestions. This work is supported by the Natural Science Foundation of Fujian Province (No.2020J05058) and the Fundamental Research Funds for the Central Universities of Huaqiao University (ZQN-903).

    The authors declare no conflict of interest.



    [1] Z. F. Zhang, L. Z. Liu, J. F. Wang, Adjacent strong edge coloring of graphs, Appl. Math. Lett., 15 (2002), 623–626. https://doi.org/10.1016/S0893-9659(02)80015-5 doi: 10.1016/S0893-9659(02)80015-5
    [2] P. N. Balister, E. Györi, J. Lehel, R. H. Schelp, Adjacent vertex distinguishing edge-colorings, SIAM J. Discrete Math., 21 (2007), 237–250. https://doi.org/10.1137/S0895480102414107 doi: 10.1137/S0895480102414107
    [3] H. Hatami, Δ+300 is a bound on the adjacent vertex distinguishing edge chromatic number, J. Combin. Theory Ser. B, 95 (2005), 246–256. https://doi.org/10.1016/j.jctb.2005.04.002 doi: 10.1016/j.jctb.2005.04.002
    [4] G. Joret, W. Lochet, Progress on the adjacent vertex distinguishing edge coloring conjecture, SIAM J. Discrete Math., 34 (2020), 2221–2238. https://doi.org/10.1137/18M1200427 doi: 10.1137/18M1200427
    [5] M. Horňák, D. J. Huang, W. F. Wang, On neighbor-distinguishing index of planar graphs, J. Graph Theory, 76 (2014), 262–278. https://doi.org/10.1002/jgt.21764 doi: 10.1002/jgt.21764
    [6] X. W. Yu, C. Q. Qu, G. H. Wang, Y. Q. Wang, Adjacent vertex distinguishing colorings by sum of sparse graphs, Discrete Math., 339 (2016), 62–71. https://doi.org/10.1016/j.disc.2015.07.011 doi: 10.1016/j.disc.2015.07.011
    [7] H. Hocquard, M. Montassier, Adjacent vertex-distinguishing edge coloring of graphs with maximum degree Δ, J. Comb. Optim., 26 (2013), 152–160. https://doi.org/10.1007/s10878-011-9444-9 doi: 10.1007/s10878-011-9444-9
    [8] M. Bonamy, J. Przybyło, On the neighbor sum distinguishing index of planar graphs, J. Graph Theroy, 85 (2017), 669–690. https://doi.org/10.1002/jgt.22098 doi: 10.1002/jgt.22098
    [9] D. J. Huang, Z. K. Miao, W. F. Wang, Adjacent vertex distinguishing indices of planar graphs without 3-cycles, Discrete Math., 338 (2015), 139–148. https://doi.org/10.1016/j.disc.2014.10.010 doi: 10.1016/j.disc.2014.10.010
    [10] Y. Wang, J. Cheng, R. Luo, G. Mulley, Adjacent vertex-distinguishing edge coloring of 2-degenerate graphs, J. Comb. Optim., 31 (2016), 874–880. https://doi.org/10.1007/s10878-014-9796-z doi: 10.1007/s10878-014-9796-z
    [11] W. F. Wang, Y. Q. Wang, Adjacent vertex-distinguishing edge coloring of K4-minor free graphs, Appl. Math. Lett., 24 (2011), 2034–2037. https://doi.org/10.1016/j.aml.2011.05.038 doi: 10.1016/j.aml.2011.05.038
    [12] G. J. Chang, D. D. Liu, Strong edge-coloring for cubic Halin graphs, Discrete Math., 312 (2012), 1468–1475. https://doi.org/10.1016/j.disc.2012.01.014 doi: 10.1016/j.disc.2012.01.014
  • This article has been cited by:

    1. Ningge Huang, Yi Tan, Lily Chen, On the inclusion chromatic index of a Halin graph, 2025, 348, 0012365X, 114266, 10.1016/j.disc.2024.114266
  • Reader Comments
  • © 2023 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(1492) PDF downloads(66) Cited by(1)

Figures and Tables

Figures(6)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog