Keywords

1 Introduction

To solve the Channel Assignment Problem in wireless communication networks, Fouquet and Jolivet [8] first introduced the notion of strong edge coloring in 1983. A strong k-edge coloring of a graph G is a mapping \(c: E(G)\rightarrow \{1,2,3,\cdots ,k\}\) such that \(c(e)\ne c(e')\) for any two edges e and \(e'\) with distance at most two. The smallest integer k such that G has a strong k-edge coloring of G is called the strong chromatic index of G, written \(\chi _s'(G)\). By greedy algorithm, it is easy to see that \(2\varDelta ^{2}-2\varDelta +1\) is a trivial upper bound on \(\chi _s'(G)\), where \(\varDelta \) is the maximum degree of G. However, it is NP-complete to decide wether \(\chi _s'(G)=k\) holds for a general graph G [14]. In 1989, Erdős and Nešetřil [7] proposed the following important conjecture while studying the strong edge coloring of graphs.

Conjecture 1

[7] For any graph G with maximum degree \(\varDelta \), \(\chi _{s}'(G)\le \frac{5}{4}\varDelta ^{2}\) if \(\varDelta \) is even, \(\chi _{s}'(G)\le \frac{5}{4}\varDelta ^{2}-\frac{1}{2}\varDelta +\frac{1}{4}\) if \(\varDelta \) is odd.

In [7], Erdős and Nešetřil constructed two classes of graphs satisfying \(\chi _{s}'(G)=\chi '(G)=|E(G)|\) while |E(G)| attains the upper bound in Conjecture 1. This illustrate that the upper bound is sharp if Conjecture 1 is true. Also, they asked a question: For a general graph G, is there any positive number \(\varepsilon \) such that \(\chi _{s}'(G)\le (2-\epsilon )\varDelta ^{2}\), where \(\varDelta \) is the maximum degree of G. As yet, there are many research results on strong edge coloring. For a graph G with sufficient large \(\varDelta \), Molloy and Reed [15] proved that \(\chi _{s}'(G)\le 1.998\varDelta ^{2}\) using probabilistic methods. In the next decides, this result was improved to \(1.93\varDelta ^{2}\) by Bruhn and Joos [4], \(1.835\varDelta ^{2}\) by Bonamy, Perrett and Postle [3]. For graphs with small \(\varDelta \), scholars also made a lot of research works. It is an obvious result that \(\chi _{s}'(G)\le 5=\frac{5}{4}\varDelta ^{2}\) while \(\varDelta =2\). For subcubic graphs, the above conjecture was verified by Andersen [1], and independently by Horák, Qing, Trotter [10]. For subquartic graphs, \(\chi _{s}'(G)\le 22\) was proven by Cranston [6] using algorithms. Huang, Santana and Yu [11] reduced 22 to 21. For graphs with \(\varDelta =5\), Zang [18] confirmed that \(\chi _{s}'(G)\le 37\).

For graphs with maximum average degree restriction, there are also a mount of results. The maximum average degree of a graph G, written mad(G), is the largest average degree of its subgraph. In other words, \(mad(G)=\max \{\frac{2|E(H)|}{|V(H)|}:H\subseteq G\}\). In 2013, Hocquard [9] studied the strong chromatic index of subcubic graphs with maximum average degree and obtained the following theorem.

Theorem 1

[9] Let G be a graph with \(\varDelta (G)=3\).

  1. (1)

    If \(mad(G)<\frac{7}{3}\), then \(\chi '_{s}(G)\le 6\);

  2. (2)

    If \(mad(G)<\frac{5}{2}\), then \(\chi '_{s}(G)\le 7\);

  3. (3)

    If \(mad(G)<\frac{8}{3}\), then \(\chi '_{s}(G)\le 8\);

  4. (4)

    If \(mad(G)<\frac{20}{7}\), then \(\chi '_{s}(G)\le 9\).

The given upper bound on mad(G) in Theorem 1(1)(2)(4) is optimal since there exist subcubic graphs with \(mad(G)=\frac{7}{3}\) (or \(mad(G)=\frac{5}{2}, \frac{20}{7}\)) and \(\chi _s'(G)>6\) (or \(\chi _s'(G)>7,9\)), see Fig. 1.

Fig. 1.
figure 1

\(mad(G)=\frac{7}{3}\) (or \(\frac{5}{2}, \frac{20}{7}\)) and  \(\chi _s'(G)=7\) (or \(\chi _s'(G)=8, 10\))

For subquartic graphs with bounded maximum average degree, Lv et al. [13] gave out the following theorem, which improved the corresponding upper bound on mad(G) due to Bensmail et al. [2].

Theorem 2

[13] Let G be a graph with \(\varDelta (G)=4\).

  1. (1)

    If \(mad(G)<\frac{61}{18}\), then \(\chi _s'(G)\le 16\);

  2. (2)

    If \(mad(G)<\frac{7}{2}\), then \(\chi _s'(G)\le 17\);

  3. (3)

    If \(mad(G)<\frac{18}{5}\), then \(\chi _s'(G)\le 18\);

  4. (4)

    If \(mad(G)<\frac{26}{7}\), then \(\chi _s'(G)\le 19\);

  5. (5)

    If \(mad(G)<\frac{51}{13}\), then \(\chi _s'(G)\le 20\).

Ruksasakchai and Wang [17] studied the strong edge coloring of graphs with \(\varDelta (G)\le 4\) and \(mad(G)<3\) and obtained the following theorem.

Theorem 3

[17] If G is a graphs G with maximum degree \(\varDelta \le 4\) and \(mad(G)<3\), then \(\chi _{s}'(G)\le 3\varDelta +1\).

For graphs with maximum degree 5 and bounded maximum average degree, Qin et al. [16] obtained the following theorem.

Theorem 4

[16] Let G be a graph with \(\varDelta (G)=5\).

  1. (1)

    If \(mad(G)<\frac{8}{3}\), then \(\chi _s'(G)\le 13\);

  2. (2)

    If \(mad(G)<\frac{14}{5}\), then \(\chi _s'(G)\le 14\).

Additionally, Choi et al. [5] studied the strong edge coloring of graphs with maximum degree \(\varDelta \ge 7\) and bounded maximum average degree. They obtained a theorem as follows.

Theorem 5

[5] Let G be a graph with maximum degree \(\varDelta \).

  1. (1)

    If \(\varDelta \ge 9\) and \(mad(G)<\frac{8}{3}\), then \(\chi _s'(G)\le 3\varDelta -3\);

  2. (2)

    If \(\varDelta \ge 7\) and \(mad(G)<3\), then \(\chi _s'(G)\le 3\varDelta \).

Recently, Li et al. [12] studied the strong edge coloring of graphs with maximum degree \(\varDelta \ge 6\) and bounded maximum average degree. The following theorem is given in [12].

Theorem 6

[12] Let G be a graph with maximum degree \(\varDelta \).

  1. (1)

    If \(\varDelta \ge 6\) and \(mad(G)<\frac{23}{8}\), then \(\chi _s'(G)\le 3\varDelta -1\);

  2. (2)

    If \(\varDelta \ge 7\) and \(mad(G)<\frac{26}{9}\), then \(\chi _s'(G)\le 3\varDelta -1\).

In this paper, we further consider the strong edge coloring of subquartic graphs by using color exchange method and discharging method. We obtained the following theorem.

Theorem 7

If G is a graph with \(\varDelta (G)=4\) and \(mad(G)<\frac{8}{3}\), then \(\chi '_{s}(G)\le 11\).

Fig. 2.
figure 2

Subquartic graphs.

\(G_1, G_2, G_3\) in Fig. 2 are subquartic graphs, where \(mad(G_1)=\frac{8}{3}\), \(\chi _s'(G_1)=10\); \(mad(G_2)=\frac{20}{7}\), \(\chi _s'(G_2)=11\) and \(mad(G_3)=3\), \(\chi _s'(G_3)=12\) (we can take the graph obtained from \(G_1\) by deleting two 1-vertices, \(G_2\) by deleting the 1-vertex and \(G_3\) as subgraphs, respectively). We do not know whether the upper bound \(mad(G)<\frac{8}{3}\) in Theorem 7 is optimal. However, due to the graph \(G_3\) in Fig. 2, we know that there exists a graph G with \(\varDelta (G)=4\), \(mad(G)=3\) and \(\chi '_{s}(G)=12\).

For the strong edge coloring of subquartic graphs, Theorem 2 gives out some sufficient conditions for \(\chi '_{s}(G)\le 16\) (respectively 17,18,19,20). Theorem 3 indicates that any graph G with \(\varDelta (G)=4\) and \(mad(G)<3\) satisfies \(\chi '_{s}(G)\le 13\). Therefore, Theorem 7 enriches the results of strong edge coloring for subquartic graphs.

2 Notations

All graphs considered here are finite undirected simple graphs. For a graph G, V(G), E(G), \(\varDelta (G)\) and \(\delta (G)\) denote its vertex set, edge set, maximum degree and minimum degree respectively. For \(v\in V(G)\), \(d_G(v)\) (abbreviated by d(v)) denotes the degree of v in G. v is a i (or \(i^+\), \(i^-\))-vertex if \(d(v)=i\) (or \(d(v)\ge i\), \(d(v)\le i\)). For a vertex v, a i-neighbor of v is a i-vertex in N(v). A \(i_j\)-vertex is a i-vertex adjacent to exactly j 2-vertices. A 2-vertex is bad if it is adjacent to a 2-vertex, semi-bad if it is adjacent to a \(3_2\)-vertex. A 2-vertex is good if it is neither bad nor semi-bad. For an edge e, F(e) denotes the set of forbidden colors for it.

3 Proof of Theorem 7

Suppose G is a counterexample with minimum \(2^{+}\)-vertices and then with minimum edges. Let H be the graph obtained from G by deleting all 1-vertices. Obviously, \(H\subseteq G\) and then \(mad(H)\le mad(G)<\frac{8}{3}\). In the following, we first illustrate some properties of H.

Lemma 1

H does not have vertices of degree 1.

Proof

Suppose v is a 1-vertex in H and \(uv\in E(H)\). Since H is the graph obtained from obtained from G by deleting all 1-vertices, \(d_G(v)>1\) and v has at least one 1-neighbor \(v_1\) in G. Compared with G, \(G-v_1\) has the same \(2^{+}\)-vertices but fewer edges. By the minimality of G, \(\chi _s'(G-v_1)\le 11\). Note that in G, \(|F(vv_1)|\le 6\). Thus, \(vv_1\) can be colored, which leads to a contradiction.

Lemma 2

If \(d_H(v)=2\), then \(d_G(v)=2\).

Proof

Suppose \(d_G(v)>2\). Then, v has at least one 1-neighbor \(v_1\) in G. Compared with G, \(G-v_1\) has fewer edges while the same \(2^{+}\)-vertices. By the minimality of G, \(\chi _s'(G-v_1)\le 11\). Note that in G, \(|F(vv_1)|\le 9\). Thus, \(vv_1\) can be colored, which leads to a contradiction.

Lemma 3

If v is a \(3_i\)-vertex in H, where \(i\ge 1\), then \(d_G(v)=3\).

Proof

Suppose \(d_G(v)>3\). Then, v has at least one 1-neighbor \(v'\) in G. Let \(v_1\) be a 2-neighbor of v in H, By Lemma 2, \(d_G(v_1)=2\). Let \(G'=G-v'\). Compared with G, \(G'\) has the same \(2^{+}\)-vertices but fewer edges. By the minimality of G, \(\chi _s'(G-v_1)\le 11\). Note that in G, \(|F(vv')|\le 10\). Thus, \(vv'\) can be colored, which leads to a contradiction.

Lemma 4

Every bad vertex in H is adjacent to a 4-vertex.

Proof

Suppose v is a bad vertex in H and it is adjacent to a 2-vertex u and a \(3^{-}\)-vertex w. By Lemma 2, \(d_G(u)=d_G(v)=2\). Denote \(N_G(u)=\{u_1,v\}\). Note that \(2\le d_H(w)\le 3\). If \(d_H(w)=2\), then \(d_G(w)=2\) by Lemma 2. If \(d_H(w)=3\), then by Lemma 3, \(d_G(w)=3\) since \(d_H(v)=2\). Let \(G'=G-uv+ww_1\), where \(ww_1\) is a pendent edge incident with w. Note that \(3\le d_{G'}(w)\le 4\) and \(G'\) has fewer \(2^{+}\)-vertices than G. By the definition of maximum average degree, \(mad(G')<2\) if \(mad(G)<2\) and \(mad(G')\le mad(G)<\frac{8}{3}\) if \(2\le mad(G)<\frac{8}{3}\). By the minimality of G, \(\chi _s'(G')\le 11\). Let c be a strong 11-edge coloring of \(G'\). Note that in G, \(|F(uv)|\le 8\). If \(c(uu_1)\ne c(vw)\), then uv can be colored, which is a contradiction. If \(c(uu_1)=c(vw)\), then we first exchange the colors on pendant edges wv and \(ww_1\) in \(G'\). After that, uv can be colored, which leads to a contradiction.

Fig. 3.
figure 3

.

Fig. 4.
figure 4

.

Fig. 5.
figure 5

.

Lemma 5

H does not have \(3_3\)-vertices.

Proof

Suppose v is a \(3_3\)-vertex in H and \(N_H(v)=\{v_1,v_2,v_3\}\). By Lemma 2, \(d_G(v_i)=2\), \(i=1,2,3\). By Lemma 3, \(d_G(v)=3\). Let \(G'=G-v\). Note that \(G'\) has fewer \(2^{+}\)-vertices than G. By the minimality of G, \(\chi _s'(G')\le 11\). Note that in G, \(|F(vv_i)|\le 6\), \(i=1,2,3\), \(vv_i\) can be colored, which is a contradiction.

Lemma 6

Every semi-bad vertex in H is adjacent to a 4-vertex.

Proof

Suppose v is a semi-bad vertex in H and it is adjacent to a \(3_2\)-vertex u and a \(3^{-}\)-vertex w. Let \(N_H(u)=\{u_1,u_2,v\}\), where \(d_H(u_1)=2\) (see Fig. 3). By Lemma 2, \(d_G(u_1)=d_G(v)=2\). By Lemma 3, \(d_G(u)=3\). Note that \(2\le d_H(w)\le 3\) and \(d_H(v)=2\), we have \(d_G(w)=d_H(w)\) by Lemma 2 and Lemma 3. Let \(G'=G-uv+ww_1\), where \(ww_1\) is a pendant edge incident with w. Note that \(G'\) has fewer \(2^{+}\)-vertices than G, by the definition of maximum average degree, \(mad(G')<2\) if \(mad(G)<2\) and \(mad(G')\le mad(G)<\frac{8}{3}\) if \(2\le mad(G)<\frac{8}{3}\). By the minimality of G, \(\chi _s'(G')\le 11\). Let c be a strong 11-edge coloring of \(G'\). Erase on color on \(uu_1\). Note that in G, \(|F(uu_1)|\le 9\), \(|F(uv)|\le 9\). If \(c(uu_2)\ne c(vw)\), then \(uu_1, uv\) can be colored, which is a contradiction. If \(c(uu_2)=c(vw)\), then we first exchange the colors on pendant edges wv and \(ww_1\) in \(G'\). After that, \(uu_1\) and uv can be colored, which leads to a contradiction.

Lemma 7

Let v be a \(4_i\)-vertex in H, where \(i\ge 3\). Then its 2-neighbors are all good vertices.

Proof

Suppose that \(v_1, v_2, v_3\) are 2-neighbors of v and at least one of them is not good. Without loss of generality, we assume that \(v_1\) is not a good vertex. This implies that \(v_1\) is adjacent to a 2-vertex or a \(3_2\)-vertex.

If \(v_1\) is adjacent to a 2-vertex u (see Fig. 4), then by Lemma 2, \(d_G(v_i)=d_G(u)=2\), \(i=1,2,3\). Let \(G'=G-v_1\). Note that \(G'\) has fewer \(2^{+}\)-vertices than G. By the minimality of G, \(\chi _s'(G')\le 11\). Note that in G, \(|F(uv_1)|\le 7\), \(|F(vv_1)|\le 9\), \(uv_1, vv_1\) can be colored, which is a contradiction.

If \(v_1\) is adjacent to a \(3_2\)-vertex \(v_1'\) and \(u\ne v_1\) is the other 2-neighbor of \(v_1'\) (see Fig. 5). By Lemma 2, \(d_G(v_i)=d_G(u)=2\), \(i=1,2,3\). By Lemma 3, \(d_G(v_1')=3\). Let \(G'=G-v_1\). Note that \(G-v_1\) has fewer \(2^{+}\)-vertices than G. By the minimality of G, \(\chi _s'(G')\le 11\). Note that in G, \(|F(v_1v_1')|\le 9\), \(|F(vv_1)|\le 10\). Thus, \(vv_1, v_1v_1'\) can be colored in order, which is a contradiction.

Proof of Theorem 7:  We define weight function \(w(v)=d(v)\) for each \(v\in V(H)\) and we define five discharging rules R1-R5 as follows. Let \(w'(v)\) be the final weight function while discharging finished. As we know, the sum weigh is fixed. However, we shall prove that \(w'(v)\ge \frac{8}{3}\) for each \(v\in V(H)\). This will lead to a contradiction as follow.

$$\frac{8}{3}|V(H)|\le \sum \limits _{v\in V(H)} w'(v) =\sum \limits _{v\in V(H)} w(v)\le mad(H)|V(H)|<\frac{8}{3}|V(H)|.$$

Discharging Rules:

R1 Each 4-vertex gives \(\frac{2}{3}\) to each adjacent bad vertex.

R2 Each 4-vertex gives \(\frac{1}{2}\) to each adjacent semi-bad vertex.

R3 Each 4-vertex gives \(\frac{1}{3}\) to each adjacent good vertex.

R4 Each \(3_2\)-vertex gives \(\frac{1}{6}\) to each adjacent semi-bad vertex.

R5 Each \(3_1\)-vertex gives \(\frac{1}{3}\) to each adjacent good vertex.

In the following, we shall verify that \(w'(v)\ge \frac{8}{3}\) for each \(v\in V(H)\).

By Lemma 1, \(\delta (H)\ge 2\).

  • \(d(v)=2\)

If v is bad, then by Lemma 4, v is adjacent to a 4-vertex. By R1, \(w'(v)=2+\frac{2}{3}=\frac{8}{3}\).

If v is semi-bad, then by Lemma 6, v is adjacent to a 4-vertex. By R2 and R4, \(w'(v)=2+\frac{1}{2}+\frac{1}{6}=\frac{8}{3}\).

If v is good, then by the definition of good vertex and Lemma 5, each neighbor of v is either \(3_1\)-vertex or 4-vertex. By R3 and R5, \(w'(v)=2+\frac{1}{3}\times 2=\frac{8}{3}\).

  • \(d(v)=3\)

By Lemma 5, v is \(3_i\)-vertex, where \(0\le i\le 2\).

If v is a \(3_2\)-vertex, then by R4, \(w'(v)\ge 3-\frac{1}{6}\times 2=\frac{8}{3}\).

If v is a \(3_1\)-vertex, then by R5, \(w'(v)\ge 3-\frac{1}{3}=\frac{8}{3}\).

If v is a \(3_0\)-vertex, then \(w'(v)=w(v)=3\).

  • \(d(v)=4\)

If v is a \(4_i\)-vertex, where \(i\ge 3\), then by Lemma 7, the 2-neighbors of v are good. Thus, \(w'(v)\ge 4-\frac{1}{3}\times 4=\frac{8}{3}\) by R3.

If v is a \(4_i\)-vertex, where \(0\le i\le 2\), then by R1-R3, \(w'(v)\ge 4-\frac{2}{3}\times 2=\frac{8}{3}\).

Therefore, for each \(v\in V(H)\), \(w'(v)\ge \frac{8}{3}\) and the proof of Theorem 7 is finished.   \(\square \)

4 Further Considered Problems

Theorem 7 illustrates that \(\chi '_{s}(G)\le 11\) holds for any graph G with \(\varDelta (G)=4\) and \(mad(G)<\frac{8}{3}\). For the graph \(G_3\) in Fig. 2, it satisfies that \(\varDelta (G_3)=4\), \(mad(G_3)=3\) and \(\chi '_{s}(G_3)=12\). Then a question follows out naturally. What is the supremum M such that any graph G with \(\varDelta (G)=4\) and \(mad(G)< M\) satisfying \(\chi '_{s}(G)\le 11\)?