
An acyclic edge coloring of a graph G is a proper edge coloring such that no bichromatic cycles are produced. The acyclic chromatic index of G, denoted by χ′a(G), is the smallest integer k such that G is acyclically edge k-colorable. In this paper, we consider the planar graphs without 3-cycles and intersecting 4-cycles, and prove that χ′a(G)≤Δ(G)+1 if Δ(G)≥8.
Citation: Yuehua Bu, Qi Jia, Hongguo Zhu, Junlei Zhu. Acyclic edge coloring of planar graphs[J]. AIMS Mathematics, 2022, 7(6): 10828-10841. doi: 10.3934/math.2022605
[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] | 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 |
[4] | 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 |
[5] | 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 |
[6] | Yueying Zhao, Lianying Miao . Planar graphs without $ \{4, 6, 8\} $-cycles are 3-choosable. AIMS Mathematics, 2021, 6(9): 10253-10265. doi: 10.3934/math.2021593 |
[7] | Yuehua Bu, Qiang Yang, Junlei Zhu, Hongguo Zhu . Injective coloring of planar graphs with girth 5. AIMS Mathematics, 2023, 8(7): 17081-17090. doi: 10.3934/math.2023872 |
[8] | Yanyi Li, Lily Chen . Injective edge coloring of generalized Petersen graphs. AIMS Mathematics, 2021, 6(8): 7929-7943. doi: 10.3934/math.2021460 |
[9] | Yanping Yang, Yang Wang, Juan Liu . List vertex arboricity of planar graphs without 5-cycles intersecting with 6-cycles. AIMS Mathematics, 2021, 6(9): 9757-9769. doi: 10.3934/math.2021567 |
[10] | Ningge Huang, Lily Chen . AVD edge-colorings of cubic Halin graphs. AIMS Mathematics, 2023, 8(11): 27820-27839. doi: 10.3934/math.20231423 |
An acyclic edge coloring of a graph G is a proper edge coloring such that no bichromatic cycles are produced. The acyclic chromatic index of G, denoted by χ′a(G), is the smallest integer k such that G is acyclically edge k-colorable. In this paper, we consider the planar graphs without 3-cycles and intersecting 4-cycles, and prove that χ′a(G)≤Δ(G)+1 if Δ(G)≥8.
All graphs considered in this paper are finite simple graphs. For a graph G, we use V(G), E(G), F(G), Δ(G) (Δ for short and reserved), and g(G) to denote the vertex set, edge set, face set, maximum degree and girth, respectively. A graph G is 2-connected if there are two paths between any two distinct vertices.
Let G be a planar graph. The acyclic edge k-coloring of graph G is a mapping c:E(G)→{1,2,…,k} such that any two adjacent edges receive different colors, and there are no bichromatic cycles in G. The acyclic chromatic index of G, denoted by χ′a(G), is the smallest integer k such that G is acyclically edge k-colorable.
Fiamˇcik posed a famous conjecture for acyclic edge coloring of any graphs.
Conjecture 1.1. [3] For any graph G, χ′a(G)≤Δ(G)+2.
The conjecture is still open.
For any graph G, Alon, McDiarmid and Reed [1] proved that χ′a(G)≤64Δ. Molloy and Reed [8] improved this bound to 16Δ. Later, Fialho et al. [4] showed that χ′a(G)≤3.569(Δ−1), and most recently to 2Δ−1 by Kirousis and Livieratos [7].
There have been numerous investigations about acyclic edge coloring of planar graphs.
For any planar graph G, Basavaraju et al. [2] proved that χ′a(G)≤Δ+12, Wang et al. [9] proved that χ′a(G)≤Δ+7, Wang and Zhang [10] proved that χ′a(G)≤Δ+6.
Let G be a planar graph with small grith. Shu, Wang and Wang [11] proved that χ′a(G)≤Δ(G)+2 if g(G)≥4. For planar graph G with g(G)≥5, Hou et al.[6] proved that χ′a(G)≤Δ(G)+1; they also proved that such graph has χ′a(G)=Δ(G) if Δ(G)≥9. For planar graph G without 4-cycles, Wang and Sheng [13] proved that χ′a(G)≤Δ(G)+3. Then Wang, Shu and Wang [14] improved this bound to Δ(G)+2 when Δ(G)≥5. In 2012, Fiedorowicz [5] proved that the planar graph G without an i-cycle intersect to a j-cycle has χ′a(G)≤Δ(G)+2 for i,j∈{3,4}. Most recently, Shu et al. [12] proved that the planar graph G without intersecting triangles has χ′a(G)≤Δ(G)+2.
In this paper, we consider the planar graph without 3-cycles and intersecting 4-cycles, and prove the following theorem:
Theorem 1.1. Let G be a planar graph without 3-cycles and intersecting 4-cycles. If Δ(G)≥8, then χ′a(G)≤Δ(G)+1.
Let G be a simple planar graph. For a vertex v∈V(G), N(v) denotes the set of vertices adjacent to v, and d(v)=|N(v)| denotes the degree of v. For f∈F(G), we use b(f) to denote the boundary walk of f and write f=[u1u2…un] if u1,u2,…,un are the vertices on b(f) enumerated in the clockwise direction. For f=[u1u2…un], let δ(f) denote the minimum degree of any vertex on b(f). That is, δ(f)=min{d(ui),i=1,…,n}. The degree of a face f, denoted by d(f), is the number of edges in its boundary walk.
For f∈F(G), f is called a k-(or k+-, or k−-) face if d(f)=k (or d(f)≥k, or d(f)≤k). For v∈V(G), v is called a k-(or k+-, or k−-) vertex if d(v)=k (or d(v)≥k, or d(v)≤k). If u∈N(v) and d(u)=k, then u is called k-neighbor of v. Let Nk(v)={x∈N(v)|d(x)=k}, and nk(v)=|Nk(v)|.
Let c be an edge coloring of G and v be a vertex of G. Then, C(v)={c(uv):u∈N(v)}, Fcv(uv)=C(v)∖{c(uv)}. Let α,β be two colors. An (α,β)-bichromatic path with respect to c is a path consisting of edges that are colored with α and β alternately. An (α,β)-bichromatic path which starts at the vertex u via an edge colored α and ends at v via an edge colored α is an (α,β)(u,v)-bichromatic path. We use "w.l.o.g." as a shorthand for "without loss of generality".
We apply a discharging procedure to prove Theorem 1.1. Discharging is a tool in a two-pronged approach to inductive proofs. It can be viewed as an amortized counting argument used to prove that a global hypothesis guarantees the existence of some desirable local configurations. In an application of the resulting structure theorem, one shows that each such local configuration cannot occur in a minimal counterexample to the desired conclusion. Such local configurations are called reducible configurations. In this section, we give some reducible configurations.
Let G be a counterexample with minimum |V(G)|+|E(G)| of Theorem 1.1. In other words, G is a connected simple planar graph without 3-cycles and intersecting 4-cycles, Δ=Δ(G)≥8, but χ′a(G)≥Δ+2. Let C be a color set of G, C={1,2,…,Δ+1}. Now, we discuss the structures of G.
Lemma 3.1. The graph G is 2-connected.
Proof. By contradiction, suppose that v is a cut vertex of G. Let C1,C2,…,Ct(t≥2) be the connected components of G∖v. For each 1≤i≤t, there is an acyclic (Δ+1)-edge coloring ci of Gi=Ci∪{v}. We can adjust the colors in each ci such that the colors appearing on the edges incident with v are all distinct. Now the union of these colorings is an acyclic (Δ+1)-edge coloring of G, a contradiction.
Lemma 3.2. The graph G does not contain a 2-vertex adjacent to a 3−-vertex.
Proof. By contradiction, let d(v)=2,N(v)={u,w}, d(u)≤3 (See Figure 1(1)). We prove the case d(u)=3, and d(u)=2 can be proved in a similar way. Let N(u)={v,u1,u2}, G′=G−uv. By the minimality of G, G′ admits an acyclic (Δ+1)-edge coloring c. Suppose that c(uui)=i for i=1,2.
We consider the following two cases.
Case 3.1. |C(u)∩C(v)|=0.
Since |C∖(C(u)∪C(v))|=Δ+1−3=Δ−2>0, we can color uv with α for α∈C∖(C(u)∪C(v)). Therefore, c can be extended to be an acyclic (Δ+1)-edge coloring of G, a contradiction.
Case 3.2. |C(u)∩C(v)|=1.W.l.o.g., assume that c(vw)=c(uu1)=1. If there exists a color γ∈{3,…,Δ+1} such that G contains no (1,γ)(u,v)-bichromatic path via u1 and w, then we can color uv with γ. In this way G is acyclically edge (Δ+1)-colorable, a contradiction. Thus, there must exist a (1,α)(u,v)-bichromatic path via u1 and w for each α∈{3,4,…,Δ+1}. So C(u1)=C(w)={1,3,4,…,Δ+1}. Now we recolor vw with 2. Similarly, there must exist a (2,α)(u,v)-bichromatic path via u2 and w for each α∈{3,4,…,Δ+1}. So C(u2)={2,3,4,…,Δ+1}. Now we exchange the colors between uu1 and uu2, color uv with 3, color vw with 1. Therefore, c can be extended to be an acyclic (Δ+1)-edge coloring of G, a contradiction.
Next we discuss the degree sum of the neighbors of a particular vertex.
Lemma 3.3. Let d(v)=k∈{2,3} and N(v)={vi,1≤i≤k}. Then ∑ki=1d(vi)≥Δ+k+1.
Proof. By contradiction, assume that ∑ki=1d(vi)≤Δ+k. Let G′=G−vv1, by the minimality of G, G′ admits an acyclic (Δ+1)-edge coloring c. Since the neighbors of each 2-vertex are both 4+-vertices, we have 3≤d(vi)≤Δ+k−6 when 1≤i≤3 and k=3. Suppose that c(vvi)=i−1 for i=2,…,k. We consider the following three cases.
Case 3.3. |C(v1)∩C(v)|=0.
Since |C∖(C(v1)∪C(v))|=Δ+1−(d(v1)−1)−(d(v)−1)=Δ+3−d(v1)−d(v)≥Δ−d(v1)>0, we can color vv1 with α for α∈C∖(C(v1)∪C(v)). Therefore, c can be extended to be an acyclic (Δ+1)-edge coloring of G, a contradiction.
Case 3.4. |C(v1)∩C(v)|=1. W.l.o.g., assume that 1∈C(v1).
(i) k=2.
Since |C∖(C(v1)∪C(v2))|≥Δ+1−(d(v1)−1)−(d(v2)−1)=Δ+3−(d(v1)+d(v2))≥Δ+3−(Δ+2)=1, we can color vv1 with β for β∈C∖(C(v1)∪C(v2)).Therefore, c can be extended to be an acyclic (Δ+1)-edge coloring of G, a contradiction.
(ii) k=3.
Since |C∖(C(v1)∪C(v)∪C(v2))|≥Δ+1−(d(v1)−1)−1−(d(v2)−1)=Δ−(d(v1)+d(v2))+2≥Δ−Δ+2=2, we can color vv1 with α for α∈C∖(C(v1)∪C(v)∪C(v2)), c is an acyclic (Δ+1)-edge coloring of G, a contradiction.
Case 3.5. |C(v1)∩C(v)|=2. Namely, d(v)=k=3.
Since |C∖(C(v1)∪C(v2)∪C(v3))|≥Δ+1−(d(v1)−1)−(d(v2)−1)−(d(v3)−1)=Δ−(d(v1)+d(v2)+d(v3))+4≥1, we can color vv1 with α for α∈C∖(C(v1)∪C(v2)∪C(v3)), c is an acyclic (Δ+1)-edge coloring of G, a contradiction.
For a 2-vertex, the number of 3−-vertex and Δ-vertex of its neighbors is discussed below.
Lemma 3.4. For a 2-vertex v, let N(v)={x,y}. Then (1) n2(x)≤d(x)+d(y)−Δ−2; (2) n2(x)+n3(x)≤d(x)+d(y)−Δ−1; (3) n2(x)+n3(x)≤d(x)+d(y)−Δ−2 if d(x)<Δ.
Proof. Assume that d(x)=s+1,d(y)=t+1,N(x)={v,x1,x2,…,xs},N(y)={v,y1,y2,…,yt} (See Figure 1(2)). Let G′=G−vy, G′ admits an acyclic (Δ+1)-edge coloring c. Let c(yyi)=i for 1≤i≤t,T=C∖Fcx(vx). It is clear that |T|=Δ+1−s≥2.
If T∖Fcy(vy)≠∅, then we can recolor vx with α for α∈T∖Fcy(vy), color vy with β for β∈{t+1,…,Δ+1}(β≠α). Now c can be extended to be an acyclic (Δ+1)-edge coloring of G, a contradiction. So T⊆Fcy(vy)={1,2,…,t}.
If there are i0∈T and j0∈{t+1,…,Δ+1}, such that G contains no (i0,j0)(v,y)-bichromatic path through x, then we can color vx with i0, color vy with j0, G is acyclically edge (Δ+1)-colorable, a contradiction. So G contains (α0,j)(v,y)-bichromatic path for c(vx)=α0,α0∈T and j∈{t+1,…,Δ+1}. This means that {t+1,…,Δ+1}⊆Fcx(vx). W.l.o.g., let c(xxi)=t+i for 1≤i≤Δ+1−t. Now recolor c(vx)=i for any i∈T. Similarly, G contains (i,j)(v,y)-bichromatic path for any j∈{t+1,…,Δ+1}. So T⊆Fcxl(xxl), which means that d(xl)≥|T|+1 for 1≤l≤Δ+1−t.
(1) Since d(x)≤Δ, |T|=Δ+1−s=Δ+1−(d(x)−1)=Δ−d(x)+2≥2. This means that d(xi)≥3 for 1≤i≤Δ+1−t. Therefore, n2(x)≤d(x)−(Δ+1−t)=d(x)−Δ−1+d(y)−1=d(x)+d(y)−Δ−2.
(2) If there are at least two 3-vertices in {x1,…,xΔ+1−t}, then assume that d(x1)=d(x2)=3. Since T⊆Fcxl(xxl)(1≤l≤Δ+1−t), and d(x1)=d(x2)=3, we have T={1,2}. So Fcx1(xx1)=Fcx2(xx2)={1,2}. Now we exchange the colors between xx1 and xx2, color vx with 1, color vy with t+1, G is acyclically edge (Δ+1)-colorable, a contradiction. Therefore, n2(x)+n3(x)≤d(x)−(Δ+1−t)+1=d(x)+d(y)−Δ−1.
(3) If d(x)<Δ, then |T|=Δ+1−s=Δ+1−(d(x)−1)=Δ−d(x)+2≥3. This implies that d(xl)≥|T|+1≥4 for 1≤l≤Δ+1−t. Therefore, n2(x)+n3(x)≤d(x)−(Δ+1−t)=d(x)−Δ−1+d(y)−1=d(x)+d(y)−Δ−2.
Lemma 3.5. For 2-vertex v, let N(v)={x,y}. If n2(x)=d(x)−2, then nΔ(x)=2, n2(y)=1, n3(y)≤1, d(y)=Δ.
Proof. Let N(x)={v,x1,x2,…,xs},N(y)={v,y1,y2,…,yt}, d(xi)=2, N(xi)={x,x′i} for 3≤i≤s (See Figure 2(3)). Observe that n2(x)≤d(x)+d(y)−Δ−2 by Lemma 3.4. If n2(x)=d(x)−2, then d(y)=Δ, which means that t=Δ−1. Let G′=G−xv, G′ admits an acyclic (Δ+1)-edge coloring c. Suppose that c(xxi)=i for 1≤i≤s,T=C∖Fcy(vy). It is clear that |T|=Δ+1−(d(y)−1)=2.
(1) nΔ(x)=2,n2(y)=1.
We consider the following two cases.
Case 3.6. T∖Fcx(vx)≠∅.
We can recolor vy with α for α∈T∖Fcx(vx), color vx with β for β∈{s+1,…,Δ+1}(β≠α). Now c can be extended to be an acyclic (Δ+1)-edge coloring of G, a contradiction.
Case 3.7. T∖Fcx(vx)=∅.
That is, T⊆Fcx(vx)={1,2,…,s}.
(i) T∩{3,…,s}≠∅.
Let α∈T∩{3,…,s}. Since |C∖(C(x)∪C(xα))|≥Δ+1−(d(x)−1)−1=Δ+1−d(x)≥1, we can color vy with α, color vx with β for β∈C∖(C(x)∪C(xα)), c can be extended to be an acyclic (Δ+1)-edge coloring of G, a contradiction.
(ii) T∩{3,…,s}=∅.
That is, T = {1,2} and Fcy(vy) = {3,4,…,Δ+1}. Note that c(vy)∈{1,2}, w.l.o.g., suppose that c(vy) = 1. If there is a color γ∈{s+1,…,Δ+1} such that G contains no (1,γ)(x,v)-bichromatic path via x1 and y, then we can color vx with γ. In this way G is acyclically edge (Δ+1)-colorable, a contradiction. So there must exist a (1,j)(x,v)-bichromatic path via x1 and y for each j(j∈{s+1,…,Δ+1}), which implies that {s+1,…,Δ+1}⊆C(x1)∩C(y). Now we recolor vy with 2, and don't change the colors of the other edges, c is still an acyclic (Δ+1)-edge coloring of G−xv. As the same argument above, we have that {s+1,…,Δ+1}⊆C(x2). If there is i∈{3,4,…,s}, such that G contains no (1,i)(x,v)-bichromatic path via x1 and y, then we can color vx with i. Don't change the colors of the other edges, and remove the color of xxi, c is still an acyclic (Δ+1)-edge coloring of G−xxi, now we consider the color of xxi. W.l.o.g., assume that i=3.
If c(x3x′3)=j≥s+1, then |C(x)∪C(x3)|≤s+1=d(x)≤Δ, we can color xx3 with α for α∉(C(x)∪C(x3)), G is acyclically edge (Δ+1)-colorable, a contradiction.
If c(x3x′3)=j∈{1,2}, then it can be seen from the above argument that for each α∈{s+1,…,Δ+1}, there are (i,α)(x,v)-bichromatic paths (i=1,2). So G−xx3 contains no (i,α)(x,x3)-bichromatic path (i=1,2) for each α∈{s+1,…,Δ+1}, we can color xx3 with α, G is acyclically edge (Δ+1)-colorable, a contradiction.
If c(x3x′3)=j=3, then we can color xx3 with α for α∉(C(x)∪C(v)) since |C(x)∪C(v)|≤s+1=d(x)<Δ+1, G is acyclically edge (Δ+1)-colorable, a contradiction.
If c(x3x′3)=j∈{4,…,s}, then we can recolor xx3 with α for α∉(C(x)∪C(xj)) since |C(x)∪C(xj)|≤s+1=d(x)<Δ+1, G is acyclically edge (Δ+1)-colorable, a contradiction. Hence, there are (1,i)(x,v)-bichromatic paths via x1 and y for each i∈{3,4,…,s}, which implies that {1,3,4,…,s}⊆C(x1).
In conclusion, C(x1)=C(y)={1,3,4,…,Δ+1}. Now recolor vy with 2, and don't change the colors of the other edges, c is still an acyclic (Δ+1)-edge coloring of G−xv. As the same argument above, we have that C(x2)={2,3,4,…,Δ+1}. It follows that Fcx1(xx1)=Fcx2(xx2)=Fcy(vy)={3,4,…,Δ+1} and {1,2}⊆Fcyi(yyi) for 1≤i≤t. Then d(x1)=d(x2)=Δ+1−2+1=Δ,n2(y)=1.
(2) n3(y)≤1.
By contradiction, assume that d(y1)=d(y2)=3. Now exchange the colors between yy1 and yy2, and don't change the colors of the other edges, G−xv has a new acyclic (Δ+1)-edge coloring ϕ. Let ϕ(yy1)=α.
If α∈{s+1,…,Δ+1}, then let ϕ(xv)=α. By symmetry, suppose that ϕ(vy)=1, G is acyclically edge (Δ+1)-colorable, a contradiction.
If α∈{3,4,…,s}, assume that α=3, then let ϕ(xv)=3, remove the color of xx3, G−xx3 admits an acyclic (Δ+1)-edge coloring ϕ′. Let ϕ′(x3x′3)=β. It follows from the above argument that there is a (1,Δ)(x,v)-bichromatic path through x1, so G−xx3 contains no (1,Δ)(x,x3)-bichromatic path via x1 and x′3. Similarly, G−xx3 contains no (2,Δ)(x,x3)-bichromatic path via x2 and x′3. If β∈{1,2}, then let ϕ′(xx3)=Δ, G is acyclically edge (Δ+1)-colorable, a contradiction. If β∈{4,…,s}, then let ϕ′(xx3)=γ for γ∉(ϕ′(x)∪ϕ′(xβ)), G is acyclically edge (Δ+1)-colorable, a contradiction. If β≥s+1, then let ϕ′(xx3)=γ for γ∉(ϕ′(x)∪ϕ′(x3)), G is acyclically edge (Δ+1)-colorable, a contradiction. This implies that n3(y)≤1.
Lemma 3.6. For 2-vertex v, let N(v)={x,y}. If d(x)=4 and n2(x)+n3(x)=2, then nΔ(x)=2, n2(y)=1,n3(y)≤1,d(y)=Δ.
Proof. Note that n2(x)+n3(x)≤d(x)+d(y)−Δ−2=d(y)−Δ+2 by Lemma 3.4. It is clear that d(y)=Δ since n2(x)+n3(x)=2. Let N(x)={v,x1,x2,x3},N(y)={v,y1,y2,…,yΔ−1} (See Figure 2(4)). If n2(x)=d(x)−2=2, then nΔ(x)=2,n2(y)=1 and n3(y)≤1 by Lemma 3.5. We prove the case n2(x)=1 and n3(x)=1. Suppose that d(x3)=3. Let G′=G−xv, by the minimality of G, G′ admits an acyclic (Δ+1)-edge coloring c. Assume that c(xxi)=i for 1≤i≤3, T=C∖Fcy(vy). Clearly, |T|=Δ+1−(Δ−1)=2.
(1) nΔ(x)=2,n2(y)=1.
We consider the following two cases.
Case 3.8. T∖Fcx(vx)≠∅.
We can recolor vy with α for α∈T∖Fcx(vx), color vx with β for β∈{4,…,Δ+1}(β≠α), c is an acyclic (Δ+1)-edge coloring of G, a contradiction.
Case 3.9. T∖Fcx(vx)=∅.
That is, T⊆Fcx(vx)={1,2,3}. Since |T|=Δ+1−(Δ−1)=2, {1,2,3}∖T≠∅. As the same argument of Case 2 in Lemma 3.5, we have nΔ(x)=2,n2(y)=1 by setting s=3.
(2) n3(y)≤1.
By contradiction, suppose that d(y1)=d(y2)=3. It follows from the above argument that Fcy1(yy1)=Fcy2(yy2)={1,2}. By the symmetry, assume that c(vy)=1, exchange the colors between yy1 and yy2, G−xv has a new acyclic (Δ+1)-edge coloring ϕ. Let ϕ(yy1)=α.
If α∈{4,…,Δ+1}, then let ϕ(vx)=α, G is acyclically edge (Δ+1)-colorable, a contradiction.
If α=3, then let ϕ(xv)=3, and remove the color of xx3, G−xx3 has an acyclic (Δ+1)-edge coloring ϕ′. It follows from the above argument that there are (i,γ)(x,v)-bichromatic paths under c of G−xv for i=1,2 and γ∈{4,5,…,Δ+1}. So G−xx3 contains no (i,γ)(x,x3)-bichromatic path. If |ϕ′(x)∩ϕ′(x3)|=0, then |ϕ′(x)∪ϕ′(x3)|≤3+2=5<Δ+1. Let ϕ′(xx3)=β for β∈C∖(ϕ′(x)∪ϕ′(x3)), ϕ′ can be extended to G, a contradiction. If |ϕ′(x)∩ϕ′(x3)|=1, then suppose 1∈ϕ′(x3). Let ϕ′(xx3)=β for β∈{4,…,Δ+1}∖ϕ′(x3), ϕ′ can be extended to G, a contradiction. If |ϕ′(x)∩ϕ′(x3)|=2, then ϕ′(x)∩ϕ′(x3)={1,2}, we can let ϕ′(xx3)=β for β∈{4,…,Δ+1}, ϕ′ can be extended to G, a contradiction. This implies that n3(y)≤1.
Next we discuss the number of 3−-vertex in the neighbors of Δ-vertex.
Lemma 3.7. For d(v)=Δ, if n2(v)≥Δ−3, then n2(v)+n3(v)≤Δ−2.
Proof. Assume that v4∈N(v), N(v4)={v,v′4}. By Lemma 3.4, n2(v)≤d(v)+d(v′4)−Δ−2=d(v′4)−2≤Δ−2, n2(v)+n3(v)≤d(v)+d(v′4)−Δ−1=d(v′4)−1≤Δ−1; if n2(v)=Δ−2, then d(v′4)=Δ; if n2(v)+n3(v)=Δ−1, then d(v′4)=Δ. Note that nΔ(v)=2 when n2(v)=Δ−2 by Lemma 3.5. Namely n3(v)=0, n2(v)+n3(v)=Δ−2+0=Δ−2.
Otherwise n2(v)=Δ−3. By contradiction, suppose that n2(v)+n3(v)≥Δ−1. We have that n2(v)+n3(v)=Δ−1 since n2(v)+n3(v)≤Δ−1. Let N(v)={v1,v2,…,vΔ},N(vi)={v,xi} for 4≤i≤Δ. Assume that d(v2)=d(v3)=3 (See Figure 3(5)). Let G′=G−vvΔ, by the minimality of G, G′ admits an acyclic (Δ+1)-edge coloring c. Let c(vvi)=i for 1≤i≤Δ−1,T=C∖FcxΔ(vΔxΔ). Since n2(v)+n3(v)=Δ−1, it follows from the above argument that d(xΔ)=d(xΔ−1)=…=d(x4)=Δ, |T|=Δ+1−(Δ−1)=2.
We consider the following two cases.
Case 3.10. T∖C(v)≠∅.
We can recolor vΔxΔ with α for α∈T∖C(v), color vvΔ with β for β∈{Δ,Δ+1}(β≠α), c is an acyclic (Δ+1)-edge coloring of G, a contradiction.
Case 3.11. T∖C(v)=∅. That is, T⊆C(v)={1,2,…,Δ−1}.
(i) T∩{4,5,…,Δ−1}≠∅.
We can color vΔxΔ with i for i∈T∩{4,5,…,Δ−1}, color vvΔ with j for j∈{Δ,Δ+1}∖{c(vixi)}, G is acyclically edge (Δ+1)-colorable, a contradiction.
(ii) T∩{4,5,…,Δ−1}=∅.
That is, T⊆{1,2,3}. Note that c(vΔxΔ)∈T⊆{1,2,3}. When c(vΔxΔ)=1, if there exists a color γ∈{Δ,Δ+1}, such that G contains no (1,γ)(v,vΔ)-bichromatic path via xΔ and v1, then let c(vvΔ)=γ, G is acyclically edge (Δ+1)-colorable, a contradiction. Hence, G contains (1,i)(v,vΔ)-bichromatic paths via xΔ and v1 for each i∈{Δ,Δ+1}. This implies that {Δ,Δ+1}⊆C(v1)∩C(xΔ). If 2∈T, then we can recolor vΔxΔ with 2, c is still an acyclic (Δ+1)-edge coloring of G−vvΔ, the same argument shows that C(v2)={2,Δ,Δ+1}. If 3∈T, then we can recolor vΔxΔ with 3, c is still an acyclic (Δ+1)-edge coloring of G−vvΔ, the same argument shows that C(v3)={3,Δ,Δ+1}.
(a) 1∉T. Namely, T={2,3}.
It follows from the above argument that Fcv2(vv2)=Fcv3(vv3)={Δ,Δ+1}. Now exchange the colors between vv2 and vv3, we can color vΔxΔ with α for α∈T, color vvΔ with Δ, G is acyclically edge (Δ+1)-colorable, a contradiction.
(b) 1∈T. Suppose that T={1,2}.
It follows from the above argument that {Δ,Δ+1}⊆C(v1) and Fcv2(vv2)={Δ,Δ+1}. Now let c(vΔxΔ)=2,c(vvΔ)=7, remove the color of vv7, c is still an acyclic (Δ+1)-edge coloring of G−vv7. Let T′=C∖Fcx7(v7x7), it is clear that |T′|=Δ+1−(Δ−1)=2. It follows from the above argument that T′⊆{1,2,3}, and 1∈T′. Assume that c(v7x7)=1. Since there is a (1,Δ)(v,vΔ)-bichromatic path through v1, G−vv7 contains no (1,Δ)(v,v7)-bichromatic path via v1 and x7, we can color vv7 with Δ, G is acyclically edge (Δ+1)-colorable, a contradiction.
The following Lemma shows if a k-vertex(k∈{4,5,6}) has no 2-neighbor, then n3(v)<k.
Lemma 3.8. Let v be a k-vertex, with k∈{4,5,6}. If n2(v)=0, then n3(v)<k.
Proof. By contradiction, suppose that n3(v)=k. Let N(v)={x,v1,v2,v3,v4,vk−1},N(x)={v,x1,x2} and N(vi)={v,v′i,v″i}(1≤i≤k−1) (See Figure 3(8)). Let G′=G−xv, by the minimality of G, G′ admits an acyclic (Δ+1)-edge coloring c. Let c(vvi)=i for 1≤i≤k−1. We consider the following three cases.
Case 3.12. |C(x)∩C(v)|=0.
Note that |C∖(C(x)∪C(v))|=Δ+1−2−(k−1)=Δ−k>0, we can color xv with α for α∈C∖(C(x)∪C(v)), c is an acyclic (Δ+1)-edge coloring of G, a contradiction.
Case 3.13. |C(x)∩C(v)|=1.
W.l.o.g., assume that c(xx1)=c(vv1)=1,c(xx2)=k. If there exists a color γ∈{k+1,…,Δ+1} such that G contains no (1,γ)(x,v)-bichromatic path, then we can color xv with γ. In this way G is acyclically edge (Δ+1)-colorable, a contradiction. So there must exist (1,α)(x,v)-bichromatic paths for each α∈{k+1,…,Δ+1}, which implies that {k+1,…,Δ+1}⊆C(x1)∩C(v1). Thus, d(v1)≥Δ+1−k+1≥Δ−4≥4, a contradiction.
Case 3.14. |C(x)∩C(v)|=2.
W.l.o.g., assume that c(xx1)=c(vv1)=1,c(xx2)=c(vv2)=2.
(i) Δ≥9.
Since d(v1)=d(v2)=3 and Δ≥9, we have that {k,…,Δ+1}∖(C(v1)∪C(v2))≠∅. Let β∈{k,…,Δ+1}∖(C(v1)∪C(v2)). Note that G contains no (i,β)(x,v)-bichromatic path for i=1,2, we can color xv with β, c is an acyclic (Δ+1)-edge coloring of G, a contradiction.
(ii) Δ=8.
(a) k∈{4,5}.
Since d(v1)=d(v2)=3 and Δ≥8, we have that {k,…,Δ+1}∖(C(v1)∪C(v2))≠∅. Let β∈{k,…,Δ+1}∖(C(v1)∪C(v2)). Note that G contains no (i,β)(x,v)-bichromatic path for i=1,2, we can color xv with β. In this way c is an acyclic (Δ+1)-edge coloring of G, a contradiction.
(b) k=6.
If there exists a color γ∈{6,7,8,9} such that G contains no (i,γ)(x,v)-bichromatic path for i=1,2, then we can color xv with γ. In this way G is acyclically edge (Δ+1)-colorable, a contradiction. So {6,7,8,9}⊆(C(v1)∪C(v2)). Since d(v1)=d(v2)=3, we have that C(v1)∪C(v2)={1,2,6,7,8,9} and c(v1v′1)∉C(v2),c(v1v″1)∉C(v2). We can recolor vv2 with α for α∈{c(v1v′1),c(v1v″1)}, don't change the colors of the other edges, G′ has a new acyclic (Δ+1)-edge coloring c′, and |C′(x)∩C′(v)|=1. By the similar argument in case 2, we can get an acyclic (Δ+1)-edge coloring of G, a contradiction.
Note that G is a minimal counterexample to Theorem 1.1, and G is a connected planar graph. By Euler,s formula |V|+|F|−|E|=2 and the relation ∑v∈V(G)d(v)=∑f∈F(G)d(f)=2|E(G)|, we can derive the identity
∑v∈V(G)(d(v)−4)+∑f∈F(G)(d(f)−4)=−8. |
We define the initial charge function by ω(v)=d(v)−4 for v∈V(G) and ω(f)=d(f)−4 for f∈F(G). It follows from the identity that ∑x∈V(G)∪F(G)ω(x)=−8. According to the structures of G, we design some discharging rules and redistribute charge such that the total amount of charge has not changed. Once the discharging is finished, a new charge function ω′(x) is produced. Next, we prove ω′(x)≥0 for all x∈V(G)∪F(G). Therefore, we can get the following contradiction
0≤∑x∈V(G)∪F(G)ω′(x)=∑x∈V(G)∪F(G)ω(x)=−8. |
Hence, we demonstrate that the counterexample can not exist and Theorem 1.1 is proved.
Discharging rules:
(R1) Every 5+-face f sends d(f)−4d(f) to each incident vertex.
(R2) Every 4-vertex v sends 15 to each adjacent 3-vertex, and then distributes the remaining extra charge evenly among all adjacent 2-vertices.
(R3) Every 5-vertex v sends 25 to each adjacent 3-vertex, and then distributes the remaining extra charge evenly among all adjacent 2-vertices.
(R4) Every 6+-vertex v sends 35 to each adjacent 3-vertex, and then distributes the remaining extra charge evenly among all adjacent 2-vertices.
In the following, we will prove that ω′(v)≥0 for each v∈V(G). Observe that δ(G)≥2 by Lemma 3.1.
(1) d(v)=3, ω(v)=−1.
Let N(v)={v1,v2,v3}. We have that d(v1)+d(v2)+d(v3)≥Δ+4≥12 by Lemma 3.3. Since the neighbors of 2-vertex are both 4+-vertices, then n2(v)=0, that is d(vi)≥3 for i=1,2,3, and n3(v)≤2.
If n3(v)=2, suppose that d(v1)=d(v2)=3, then d(v3)≥6. This implies that ω′(v)≥−1+2×15+35 = 0 by R1,R4. If n3(v)=1, then either n4(v)=1,n5+(v)=1 or n4(v)=0,n5+(v)=2. By R1,R2,R3,R4, we have that ω′(v)≥−1+2×15+min{15+25,2×25,25+35,2×35}=0. If n3(v)=0, then n4+(v)=3. By R1,R4, ω′(v)≥−1+2×15+3×15=0.
(2) d(v)=4, ω(v)=0.
If n2(v)≠0, then n2(v)+n3(v)≤2 by Lemma 3.4.
This means that ω′(v)≥0+3×15−15n3(v)−0+3×15−15n3(v)n2(v)×n2(v)=0 by R1,R2.
If n2(v)=0, then n3(v)≤3 by Lemma 3.8, which implies that ω′(v)≥0+3×15−3×15=0 by R1,R4.
(3) d(v)=5, ω(v)=1.
If n2(v)≠0, then ω′(v)≥1+4×15−25n3(v)−1+4×15−25n3(v)n2(v)×n2(v)=0 by R1,R3.
If n2(v)=0, then n3(v)≤4 by Lemma 3.8. This implies that ω′(v)≥1+4×15−4×25=15>0 by R1,R3.
(4) d(v)=6, ω(v)=2.
If n2(v)≠0, then ω′(v)≥2+5×15−35n3(v)−2+5×15−35n3(v)n2(v)×n2(v)=0 by R1,R4.
If n2(v)=0, then n3(v)≤5 by Lemma 3.8. This implies that ω′(v)≥2+5×15−5×35=0 by R1,R4.
(5) d(v)≥7, ω(v)=d(v)−4.
If n2(v)≠0, then ω′(v)≥d(v)−4+15×(d(v)−1)−35n3(v)−d(v)−4+15×(d(v)−1)−35n3(v)n2(v)×n2(v)=0 by R1,R4.
If n2(v)=0, then ω′(v)≥d(v)−4+15×(d(v)−1)−35n3(v)≥65d(v)−215−35d(v)=35d(v)−215≥0 by R1,R4.
(6) d(v)=2, ω(v)=−2.
For convenience, let τ(u→v) denote the charge transferred out of u into v according to the above rules, u,v∈V(G). Let N(v)={x,y}, then n2(x)≤d(x)−2,n2(y)≤d(y)−2 by Lemma 3.4.
(6.1) If one of vertices in {x,y}, such as x, has n2(x) = d(x)−2, then nΔ(x) = 2, n2(y) = 1, n3(y)≤1 and d(y) = Δ by Lemma 3.5. By R1 and R4, we have that τ(y→v)≥ω(y)+15×(d(y)−1)−35n3(y)n2(y)≥d(y)−4+15d(y)−15−1×351=65d(y)−245=65×(Δ−4)≥245. This implies that ω′(v)≥−2+15+245=3>0 by R1.
(6.2) n2(x)≤d(x)−3,n2(y)≤d(y)−3. Suppose that d(x)≤d(y).
(6.2.1) d(x)≥9.
Observe that n2(x)+n3(x)≤d(x)+d(y)−Δ−1 by Lemma 3.4. Note that τ(x→v)≥ω(x)+15×(d(x)−1)−35n3(x)n2(x)≥d(x)−4+15d(x)−15−35×(d(x)−1−n2(x))n2(x)=35d(x)−185+35n2(x)n2(x)=35×(1+d(x)−6n2(x))≥35×(1+d(x)−6d(x)−3) = 35×(2−3d(x)−3) by R1,R4.
Similarly, τ(y→v)≥35×(2−3d(x)−3). Since d(x)≥9, d(x)≤d(y), we have τ(x→v)≥910,τ(y→v)≥910. This implies that ω′(v)≥−2+15+2×910=0 by R1.
(6.2.2) d(x)=8, then n2(x)≤d(x)−3=5.
(a) Δ≥9.
Observe that n2(x)+n3(x)≤d(x)+d(y)−Δ−2≤d(x)−2 by Lemma 3.4. Note that τ(x→v)≥ω(x)+15×(d(x)−1)−35n3(x)n2(x)≥d(x)−4+15×(d(x)−1)−35×(d(x)−2−n2(x))n2(x)=35×(1+d(x)−5n2(x))≥35×(1+d(x)−5d(x)−3)=35×(2−2d(x)−3)=2425 by R1,R4.
If d(y)=d(x)<Δ, then we have τ(y→v)≥35×(1+3n2(y))≥35×(1+35)=2425 by the similar argument above. This implies that ω′(v)≥−2+15+2×2425=325>0 by R4.
If d(y)≥9, then we have n2(y)+n3(y)≤d(x)+d(y)−Δ−1≤d(y)−2, n2(y)≤d(y)−3 by Lemma 3.4. This means that τ(y→v)≥35×(1+d(y)−5n2(y))≥35×(2−2d(y)−3) by R1 and R4. τ(y→v)≥1 since d(y)≥9. So ω′(v)≥−2+15+2425+1=225>0 by R4.
(b) Δ=8.
Now we have d(x)=d(y)=Δ=8. Observe that n2(x)+n3(x)≤d(x)+d(y)−Δ−1≤d(x)−1 = 7 by Lemma 3.4. If n2(x)≤4, then τ(x→v)≥35×(1+2d(x)−14n2(x))≥35×(1+24)=910 by R1,R4. Similarly, if n2(y)≤4, then τ(y→v)≥910. If n2(x)≥5, then we have n2(x)=5 due to n2(x)≤d(x)−3=5. Observe that n2(x)+n3(x)≤Δ−2=6 by Lemma 3.7. This implies that τ(x→v)≥35×(1+2d(x)−13n2(x))=35×(1+35)=2425 by R1,R4. Similarly, if n2(y)≤5, then τ(y→v)≥2425. Hence, ω′(v)≥−2+15+min{2×910,2×2425,910+2425}=0 by R1.
(6.2.3) d(x)=7, then n2(x)≤d(x)−3=4.
Observe that n2(x)+n3(x)≤d(x)+d(y)−Δ−2≤5 by Lemma 3.4. Note that τ(x→v)≥35 ×(1+2d(x)−12n2(x))≥910 by R1,R4.
If d(y)=7, then we have that n2(y)+n3(y)≤d(x)+d(y)−Δ−2=12−Δ≤4,n2(y)≤d(y)−3=4 by Lemma 3.4. Note that τ(y→v)≥ω(y)+15×(d(y)−1)−35n3(y)n2(y)≥d(y)−4+15×(d(y)−1)−35×(4−n2(y))n2(y)=35×(1+2d(y)−11n2(y))≥35 ×(1+34)=2120 by R1,R4. We have that ω′(v)≥−2+15+910+2120=320>0 by R1.
If d(y)≥8, then we have that n2(y)+n3(y)≤d(x)+d(y)−Δ−1≤d(y)−2,n2(y)≤d(y)−3 by Lemma 3.4.
Note that τ(y→v)≥ω(y)+15×(d(y)−1)−35n3(y)n2(y)≥d(y)−4+15×(d(y)−1)−35×(d(y)−2−n2(y))n2(y)≥35×(1+d(y)−5n2(y))≥35×(2−2d(y)−3) by R1,R4. Since d(y)≥8, τ(y→v)≥2425, we have that ω′(v)≥−2+15+910+2425=350>0 by R1.
(6.2.4) d(x)=6, then n2(x)≤d(x)−3=3.
By Lemma 3.4, we have that n2(x)+n3(x)≤d(x)+d(y)−Δ−2≤4. Note that τ(x→v)≥35 ×(1+2d(x)−11n2(x))≥45 by R1,R4.
If d(y)=6, then we have that n2(y)+n3(y)≤d(x)+d(y)−Δ−2=10−Δ≤2 by Lemma 3.4. So n2(y)≤2. Note that τ(y→v)≥ω(y)+15×(d(y)−1)−35n3(y)n2(y)≥d(y)−4+15×(d(y)−1)−35×(2−n2(y))n2(y)=35×(1+2d(y)−9n2(y))≥35 ×(1+32)=32 by R1,R4. This means that ω′(v)≥−2+15+45+32=12>0 by R1.
If d(y)=7, then we have that n2(y)+n3(y)≤d(x)+d(y)−Δ−2=11−Δ≤3 by Lemma 3.4. So n2(y)≤3. Note that τ(y→v)≥ω(y)+15×(d(y)−1)−35n3(y)n2(y)≥d(y)−4+15×(d(y)−1)−35×(3−n2(y))n2(y)=35×(1+2d(y)−10n2(y))≥35 ×(1+43)=75 by R1,R4. This means that ω′(v)≥−2+15+45+75=25>0 by R1.
If d(y)≥8, then we have that n2(y)+n3(y)≤d(x)+d(y)−Δ−1≤d(y)−3,n2(y)≤d(y)−3 by Lemma 3.4.
Note that τ(y→v)≥ω(y)+15×(d(y)−1)−35n3(y)n2(y)≥d(y)−4+15×(d(y)−1)−35×(d(y)−3−n2(y))n2(y)=35×(1+d(y)−4n2(y))≥35×(2−1d(y)−3) by R1,R4. Since d(y)≥8, τ(y→v)≥2725, we have that ω′(v)≥−2+15+45+2725=225>0 by R1.
(6.2.5) d(x)=5, then n2(x)≤d(x)−3=2.
By Lemma 3.4, we have that n2(x)+n3(x)≤d(x)+d(y)−Δ−2≤3. Note that τ(x→v)≥ω(x)+15×(d(x)−1)−25n3(x)n2(x)≥1+4×15−25×(3−n2(x))n2(x)=25+35×1n2(x)≥25+35×12=710 by R1,R3. Observe that d(x)+d(y)≥Δ+3 by Lemma 3.3, so d(y)≥Δ+3−d(x)≥6.
If d(y)=6, then we have that n2(y)+n3(y)≤d(x)+d(y)−Δ−2=9−Δ≤1 by Lemma 3.4, which means that n2(y)=1,n3(y)=0. Note that τ(y→v)≥ω(y)+15×(d(y)−1)−35n3(y)n2(y)=2+5×15−0×351=3 by R1,R4. So ω′(v)≥−2+15+710+3=1910>0 by R1.
If d(y)=7, then we have that n2(y)+n3(y)≤d(x)+d(y)−Δ−2=10−Δ≤2 by Lemma 3.4. So n2(y)≤2. Note that τ(y→v)≥ω(y)+15×(d(y)−1)−35n3(y)n2(y)≥d(y)−4+15×(d(y)−1)−35×(2−n2(y))n2(y)=35×(1+5n2(y))≥35 ×(1+52)=2110 by R1,R4. Hence, ω′(v)≥−2+15+710+2110=1>0 by R1.
If d(y)≥8, then we have that n2(y)+n3(y)≤d(x)+d(y)−Δ−1≤d(y)−4 by Lemma 3.4. So n2(y)≤d(y)−4. Note that τ(y→v)≥ω(y)+15×(d(y)−1)−35n3(y)n2(y)≥d(y)−4+15×(d(y)−1)−35×(d(y)−4−n2(y))n2(y)=35×(1+d(y)−3n2(y))≥35 ×(2+1d(y)−4) by R1,R4. Observe that τ(y→v)>65 since d(y)≥8. Then ω′(v)>−2+15+710+65=110>0 by R1.
(6.2.6) d(x)=4
By Lemma 3.4, we have that n2(x)+n3(x)≤d(x)+d(y)−Δ−2≤2.
(a) n2(x)+n3(x)=2.
By Lemma 3.6, we have that nΔ(x)=2,n2(y)=1,n3(y)≤1, and d(y)=Δ. Note that τ(y→v)≥ω(y)+15×(d(y)−1)−35n3(y)n2(y)≥d(y)−4+15d(y)−15−1×351=65d(y)−245=65(Δ−4)≥245 by R1,R4. So ω′(v)≥−2+15+245 = 3>0 by R1.
(b) n2(x)+n3(x)≤1. That is n2(x)=1,n3(x)=0.
Note that τ(x→v)≥ω(x)+15×(d(x)−1)−15n3(x)n2(x)≥0+3×15−0×151=35 by R1,R2. Observe that d(x)+d(y)≥Δ+3 by Lemma 3.3, so d(y)≥Δ+3−d(x)≥7.
If d(y)=7, then we have that n2(y)+n3(y)≤d(x)+d(y)−Δ−2=9−Δ≤1 by Lemma 3.4, which means that n2(y)=1,n3(y)=0. Note that τ(y→v)≥ω(y)+15×(d(y)−1)−35n3(y)n2(y)=3+6×15−0×351=215 by R1,R4. So ω′(v)≥−2+15+35+215=3>0 by R1.
If d(y)≥8, then we have that n2(y)+n3(y)≤d(x)+d(y)−Δ−1≤d(y)−5 by Lemma 3.4. Hence n2(y)≤d(y)−5.
Note that τ(y→v)≥ω(y)+15×(d(y)−1)−35n3(y)n2(y)≥d(y)−4+15×(d(y)−1)−35×(d(y)−5−n2(y))n2(y)=35×(1+d(y)−2n2(y))≥35×(2+3d(y)−5) by R1,R4. We have that τ(y→v)>65 since d(y)≥8. Hence, ω′(v)>−2+15+35+65=0 by R1.
After R1−R4, we get ω′(v)≥0, for all v∈V(G). For all f∈F(G), if d(f)=4, then ω′(f)=ω(f)=d(f)−4=0. If d(f)≥5, then we have ω′(f)=d(f)−4−d(f)−4d(f)×d(f)=0 by R4.
In summary, we get the following contradictory formula:
−8=∑x∈V(G)∪F(G)ω(x)=∑x∈V(G)∪F(G)ω′(x)≥0. |
The above contradiction indicates that G does not exist, so Theorem 1.1 is true.
In this paper, we consider the acyclic chromatic index of planar graphs without 3-cycles and intersecting 4-cycles and proved that such graphs have χ′a(G)≤Δ+1 if Δ(G)≥8. A natural problem in context of our main result is the following:
What is the optimal constant c such that χ′a(G)≤Δ(G)+1 for every planar graph G with g(G)≥c?
This work was supported by National Natural Science Foundations of China (Grant Nos. 11771403 and 11901243)
The authors declare no conflicts of interest.
[1] |
N. Alon, C. McDiarmid, B. Reed, Acyclic coloring of graphs, Random Struct. Algor., 2 (1991), 277–288. http://dx.doi.org/10.1002/rsa.3240020303 doi: 10.1002/rsa.3240020303
![]() |
[2] |
M. Basavaraju, L. Chandran, N. Cohen, F. Havet, T. M¨ulle, Acyclic edge-coloring of planar graphs, SIAM J. Discrete Math., 25 (2011), 463–478. http://dx.doi.org/10.1137/090776676 doi: 10.1137/090776676
![]() |
[3] | J. Fiamˇcik, The acyclic chromatic class of a graph(Russian), Math. Slovaca, 28 (1978), 139–145. |
[4] |
P. Fialho, B. de Lima, A. Procacci, A new bound on the acyclic edge chromatic number, Discrect Math., 343 (2020), 112037. http://dx.doi.org/10.1016/j.disc.2020.112037 doi: 10.1016/j.disc.2020.112037
![]() |
[5] |
A. Fiedorowicz, Acyclic edge colouring of plane graphs, Discrete Appl. Math., 160 (2012), 1513–1523. http://dx.doi.org/10.1016/j.dam.2012.02.018 doi: 10.1016/j.dam.2012.02.018
![]() |
[6] |
J. Hou, W. Wang, X. Zhang, Acyclic edge coloring of planar graphs with girth at least 5, Discrete Appl. Math., 161 (2013), 2958–2967. http://dx.doi.org/10.1016/j.dam.2013.06.013 doi: 10.1016/j.dam.2013.06.013
![]() |
[7] | L. Kirousis, J. Livieratos, The acyclic chromatic index is less than the double of the max degree, arXiv: 1901.07856v6. http://dx.doi.org/10.48550/arXiv.1901.07856 |
[8] |
M. Molloy, B. Reed, Further algorithmic aspects of the local lemma, Proceedings of the Thirtieth Annual ACM Symposium on the Theory of Comouting, 1998,524–529. http://dx.doi.org/10.1145/276698.276866 doi: 10.1145/276698.276866
![]() |
[9] |
W. Wang, Q. Shu, Y. Wang, A new upper bound on the acyclic chromatic indices of planar graphs, Eur. J. Combin., 34 (2013), 338–354. http://dx.doi.org/10.1016/j.ejc.2012.07.008 doi: 10.1016/j.ejc.2012.07.008
![]() |
[10] |
T. Wang, Y. Zhang, Further result on acyclic chromatic index of planar graphs, Discrete Appl. Math., 201 (2016), 228–247. http://dx.doi.org/10.1016/j.dam.2015.07.015 doi: 10.1016/j.dam.2015.07.015
![]() |
[11] |
Q. Shu, W. Wang, Y. Wang, Acyclic chromatic indices of planar graphs with girth at least 4, J. Graph Theory, 73 (2013), 386–399. http://dx.doi.org/10.1002/jgt.21683 doi: 10.1002/jgt.21683
![]() |
[12] | Q. Shu, Y. Chen, S. Han, G. Lin, E. Miyano, A. Zhang, Acyclic edge coloring conjecture is true on planar graphs without intersecting triangles, Theor. Comput. Sci., 882 (2021), 77–108. http://dx.doi.org/j.tcs.2021.06.017 |
[13] |
Y. Wang, P. Sheng, Improved upper bound for acyclic chromatic index of planar graphs without 4-cycles, J. Comb. Optim., 27 (2014), 519–529. http://dx.doi.org/10.1007/s10878-012-9524-5 doi: 10.1007/s10878-012-9524-5
![]() |
[14] |
W. Wang, Q. Shu, Y. Wang, Acyclic edge coloring of planar graphs without 4-cycles, J. Comb. Optim., 25 (2013), 562–586. http://dx.doi.org/10.1007/s10878-012-9474-y doi: 10.1007/s10878-012-9474-y
![]() |