Abstract
A theta graph is a graph obtained by joining two vertices by three internally disjoint paths of lengths 2, 1, and 2. A neighbor sum distinguishing (NSD) total coloring of G is a proper total coloring of G such that for each edge , where denotes the set of edges incident with a vertex u. In 2015, Pilśniak and Woźniak introduced this coloring and conjectured that every graph with maximum degree admits an NSD total -coloring. In this paper, we show that the listing version of this conjecture holds for any IC-planar graph with maximum degree but without theta graphs by applying the Combinatorial Nullstellensatz, which improves the result of Song et al.
1. Introduction
The graphs mentioned in this paper are finite, undirected, and simple. For undefined terminology and notations, here we follow [1]. Let be a simple graph. For a vertex , let denote the set of edges incident with u, and we use and to represent the degree and the neighborhood of u, respectively. Let (or ) and (or ) denote the maximum degree and the minimum degree of G, respectively. A t-cycle (-cycle, -cycle) is a cycle of length t (at least t, at most t). In particular, a 3-cycle with vertex set is called a -cycle. A theta graph is a graph obtained by joining two vertices by three internally disjoint paths of lengths , and .
Let k be a positive integer and . A mapping is called a proper k-total coloring of G if for any two adjacent or incident elements in . A proper k-total coloring of G is neighbor sum distinguishing (for short, NSD) if for each edge ,
The NSD total chromatic number of G, denoted by , is the smallest integer k such that G has an NSD k-total coloring. Pilśniak and Woźniak [2] posed an important conjecture in the following.
Conjecture 1
([2]). For a graph G, .
Pilśniak and Woźniak [2] showed that Conjecture 1 holds for complete graphs, bipartite graphs, cubic graphs, and 2-degenerate graphs with maximum degree . Yao et al. [3] confirmed this conjecture for any 2-degenerate graph. Li et al. [4] verified the conjecture for -minor free graphs. Yang et al. [5] proved that any planar graph with maximum degree satisfies this conjecture.
An -planar graph, introduced by Alberson [6] in 2008, is a graph that can be embedded in a plane such that each edge is crossed at most one other edge and two pairs of crossing edges share no common end vertex, i.e., two distinct crossings are independent. There are also many results about the NSD total coloring of IC-planar as follows.
Theorem 1
([7,8,9]). Conjecture 1 holds for the following families of IC-planar graphs.
(1) Any IC-planar graph with maximum degree .
(2) Every IC-planar graph with maximum degree but without 3-cycles.
(3) Any IC-planar graph with maximum degree but without theta graphs .
A k-list total assignment of G is a mapping L that assigns to each member a set of k real numbers. For a list total assignment L of G, a mapping is called an NSD total L-coloring of G if is an NSD total coloring of G and for each . The smallest integer k such that G has an NSD total L-coloring for any k-list total assignment L is called the NSD total choice number of G, denoted by . Clearly, .
There are also many results about the list version of Conjecture 1 in the following.
Conjecture 2
([2]). For a graph G, .
Obviously, Conjecture 2 implied Conjecture 1. Qu et al. [10] proved that this conjecture holds for any planar graph G with maximum degree . Wang et al. [11] confirmed Conjecture 2 for every planar graph G with maximum degree but without theta graphs . Song et al. [12] discussed the NSD total L-coloring of any IC-planar graph and obtained the following theorem.
Theorem 2
([12]). Let G be an IC-planar graph. Then, .
In this paper, we reduce the condition of (3) in Theorem 1 to and obtain the list version result as follows.
Theorem 3.
Let G be an IC-planar graph without theta graphs . Then,
2. Preliminaries
Let G be a simple graph. An ℓ-vertex (-vertex, -vertex) u of G is a vertex with (, ). We use (, ) to denote the number of ℓ-vertices (-vertices, -vertices) adjacent to u.
In 1999, Alon developed a general algebraic technique that is called Combinatorial Nullstellensatz. It has numerous applications in additive number theory, combinatorics, and graph coloring problems. In the paper, we also use Combinatorial Nullstellensatz in the following to discuss the local structure of minimal counterexample to Theorem 3.
Lemma 1
([13]). Let be an arbitrary field and with degree , where each is a natural integer. If the coefficient of the monomial in P is nonzero, and if are subsets of with , then there are such that .
Let be a positive integer and be t finite sets of real numbers. Define
Lemma 2
([10]). Let be a positive integer and be t finite sets of real numbers, where and Define by
If for , then
3. Proof of Theorem 3
Let G be a counterexample to Theorem 3 with being minimal and . For any k-list total assignment L, every subgraph of G has an NSD total L-coloring by the minimality of G. In the following, we will extend the NSD total L-coloring of to an NSD total L-coloring of G to obtain a contradiction. As is a subgraph of G, we have . Not stated otherwise, for any . Let . For the coloring , the definition of is the same as . Let X be a subset of and be a mapping. For each , let
Assume that u is a -vertex of G. For any k-list total assignment L of G, let a map satisfy the following three conditions:
- (i)
- for every ;
- (ii)
- for any two adjacent or incident members ;
- (iii)
- for any two adjacent vertices .
As follows from the assumption that and , there is a color in to color u such that the resulting coloring obtained from satisfies for each and for each . Therefore, satisfies the definition of NSD total L-coloring. Therefore, can be extend to an NSD total L-coloring of G. Thus, for simplicity, we will omit the colors of all -vertices in the following.
Theorem 3 follows from Theorem 2 if . Thus, the following Claim 1 is immediate.
Claim 1.
.
Claim 2.
There is no edge in G with and
Proof.
By contradiction, suppose that there is an edge in G with and Without loss of generality, set and Let . Then, has an NSD total L-coloring . In order to extend the coloring to an NSD total L-coloring of G, we erase the colors on u and v. Then,
Assign a variable to u, a variable to and a variable to v, respectively. Let
where and . By Appendix A, we know that By Lemma 1, there are and such that . By the definitions of P and NSD total L-coloring, we can extend to an NSD total L-coloring of G by recoloring u and v with colors and coloring with color . It is a contradiction. □
Claim 3.
There is no edge in G with and
Proof.
Suppose to be contrary that there is an edge in G with and Without loss of generality, set . Let . Then, has an NSD total L-coloring . In order to extend the coloring to an NSD total L-coloring of G, we erase the colors on u and v. Note that v is a -vertex. The color of v can be omitted when we extend to an NSD total L-coloring of G. Then,
By Lemma 2, we have that
Thus, there is a color in to color u and a color in to color such that the resulting color obtain from satisfies for each . Therefore, we can extend to an NSD total L-coloring of G. It is a contradiction. □
The proofs of the following Claims 4 and 5 are similar to the proof of Claim 2. To avoid duplication, we omit the proofs.
Claim 4.
Let u be a 6-vertex in G. Then, ; furthermore, when .
Claim 5.
There is no -cycle in G.
The proof of the following Claim 6 is similar to the proof of Claim 3. To avoid duplication, we omit the proof of Claim 6.
Claim 6.
Let u be an ℓ-vertex of G. Then, each of the following results must hold.
(1) If , then .
(2) If , then .
(3) If , then ; furthermore, when .
(4) If , then ; furthermore, when .
Let ( be the set of vertices in G of degree i (at most i, at least i) and let
Then, for each .
Claim 7.
For the graph H, each of the following results must hold.
(1) .
(2) when .
(3) when .
(4) when , when and , when and , and when .
(5) when .
Proof.
Let v be a vertex of H. Then, by the definition of H and Claim 1.
(1)–(3) By Claims 3 and 6, we have that
Thus, when and when by Equation (1). Therefore, .
(4) By Claims 4 and 6, we have that
and
Note that if and only if follows from (2) and (3). Thus, when and when and by Equation (2). As when and , when and . Therefore, when and by Claim 4. By Equation (3), it is easy to know that when .
(5) Note that when by (1)–(3). Thus, when by Claim 2. □
Claim 8.
Each 3-cycle in H is either a -cycle or a -cycle.
Proof.
Let v be a vertex of H. Then, when and when by Claim 7. Thus, there is no -cycle in H as G contains no -cycle by Claim 5. By Claim 7, we know that each -vertex is not adjacent to any -vertex in H. Therefore, each 3-cycle in H is either a -cycle or a -cycle. □
For a planar graph, we call a face a t-face (resp., a -face, a -face, an -face) if its boundary is a t-cycle (resp., a -cycle, a -cycle, an -cycle). A face is said to be incident with the vertices and edges in its boundary.
From now on, we assume that the IC-planar graph G has been embedded on a plane such that every edge is crossed by at most one other edge and the number of crossings is as small as possible. We turn all crossings of G into new 4-vertices on the plane and obtain a planar graph , which is called the associated planar graph of G. A vertex in is called a false vertex if it is not a vertex of G and real vertex otherwise. We call a face f in a false face if it is incident with one false vertex and a real face otherwise.
Let be the associated planar graph of H. For each real vertex , we use and to denote the number of real 3-faces and false 3-faces incident with v, respectively.
Claim 9.
Let v be a real vertex of . Then, each of the following results must hold.
(1) .
(2) if v is adjacent to one false 4-vertex and .
(3) if and .
(4) if and .
Proof.
(1) As G (and thus H) is an IC-planar graph without theta graphs , any two real 3-faces have no common edge in H. Thus, a real vertex v of is incident with at most real 3-faces as each 3-face contains two edges incident with v. Furthermore, statement (1) holds.
(2) By contradiction, suppose that . If , then . Thus, each edge incident with v belongs to a real 3-face incident with v since each 3-face contains two edges. Moreover, v is not adjacent to any false 4-vertex, a contradition. Therefore, .
(3) and (4) Let v be incident with two false 3-faces. Then, the two false 3-faces contain three edges incident with v since G (and thus H) is an IC-planar graph without theta graphs . Thus, the two false 3-faces cause that is reduced by 1 when and 2 when . Therefore, (3) and (4) hold. □
The discharging method, first developed in the study of the coloring of planar graphs about 100 years ago, is an important proof technique in graph theory. The method has been applied in many types of problems, especially in various graph coloring problems.
The general process of discharging is that members (usually vertices or vertices and faces) of a graph are assigned charges by certain “charging rules”, then the graph is discharged by certain “discharging rules”, during which some members get charges, and some members lose charges, while the sum of the charges keeps unchanged.
In the following, we will apply the discharging method on the associated planar graph to show that (and thus H) does not exist. Therefore, G does not exist.
Let denote the set of faces in . For each , assign the initial charge . By Euler’s formula, we have
Next, we make some discharging rules to redistribute charges among vertices and faces and keep the total charges unchanged. For simplicity, a real ℓ-vertex is still called an ℓ-vertex in the following discussion.
The discharging rules as follows:
- (R1)
- Each 3-vertex receives from each neighbor.
- (R2)
- Let z be a false 4-vertex and x be one neighbor of z in .
- (R2.1)
- When , z receives from x.
- (R2.2)
- When , z receives 1 from x.
- (R2.3)
- When , z receives from x if and from x otherwise.
- (R2.4)
- When , z receives from x if and from x otherwise.
- (R2.5)
- When , z receives from x if and from x otherwise.
- (R3)
- Each false 3-face receives 1 from the false 4-vertex incident with it.
- (R4)
- Each real 3-face receives from each 5-vertex incident with it and from -vertex incident with it.
In the following, we give a specific example about the charge change of some false 4-vertex by the discharging rules. Let u be a false 4-vertex in and with and . We take the configuration of (see Figure 1) as a specific example to illustrate how the charge of the false 4-vertex u changes by the discharging rules.
Figure 1.
A specific example about the charge change of the false 4-vertex u.
Next, we discuss the new charge of each after the discharging process. Let denote the new charge for each . Then,
In the following, we show for each to obtain a contradiction.
First, we prove for each (real or false) face . Pick arbitrarily a face z from . If z is a false 3-face, then it must be incident with a false 4-vertex. Therefore, by (R3). If z is a real 3-face, then it is easy to verify by (R4) since each real 3-face is either a -face or a -face by Claim 8. If z is a (real or false) -face, then as no rule is applied to it. Thus, for each .
Note that consists of all real vertices in and all false 4-vertices.
Second, we show for each real vertex . Choose arbitrarily a vertex z from . Since , . Note that each real vertex z gives no charge to any false 3-face and . If z is a 3-vertex, then by (R1). If z is a 4-vertex, then as each 4-vertex gives nothing away. In the following, we discuss
Note that as H is an IC-planar graph, each real vertex z is adjacent to at most a false 4-vertex and .
Let . Then, by Claim 9. Thus, by (R2) and (R4) as z is not adjacent to any 3-vertex by Claim 7.
Let . Then, by Claim 7. If z is not adjacent to any false 4-vertex, then by Claim 9. Thus, by (R1) and (R4). If z is adjacent to one false 4-vertex and , then by Claim 9. When z is not adjacent to any 3-vertex in , by (R2) and (R4). When z is adjacent to a 3-vertex in , by (R1), (R2), and (R4). If z is adjacent to one false 4-vertex and , then by Claim 9. When z is not adjacent to any 3-vertex in , by (R2) and (R4). When z is adjacent to a 3-vertex in , by (R1), (R2), and (R4).
Let . Then, by Claim 7. If , then by Claim 9. Thus, by (R1), (R2), and (R4). If , then by Claim 9. Thus, by (R1), (R2), and (R4).
Let . Then, by Claim 7. If z is not adjacent to any false 4-vertex, then by Claim 9. Thus, by (R1) and (R4). If z is adjacent to one false 4-vertex and , then by Claim 9. Thus, by (R1), (R2), and (R4). If z is adjacent to one false 4-vertex and , then by Claim 9. Thus, by (R1), (R2), and (R4).
Let . Note that each (real or false) 3-face is incident with at most one 3-vertex as two 3-vertices are not adjacent in H (and thus ) by Claim 7. If , then and by Claim 9. Thus, by (R1), (R2), and (R4). Therefore, when and when If , then and by Claim 9. Thus, by (R1), (R2), and (R4).
Therefore, for each real vertex .
Finally, we prove for each false 4-vertex . Pick arbitrarily a vertex z from . Let . Then, up to isomorphism, the induced subgraph is one of four configurations in Figure 2.
Figure 2.
Four different configurations of .
Note that z is incident with at most two false 3-faces as H is an IC-planar graph without theta graphs . By Claim 7, z is adjacent to at most two ℓ-vertices with in .
Suppose that the configuration of is in Figure 2. As z is not incident with any false 3-face and adjacent to at most two 3-vertices, it is easy to verify by (R1) and (R2).
Suppose that the configuration of is in Figure 2. Then, z is incident with one false 3-face.
Assume that z is not adjacent to any 3-vertex. Then, it is adjacent to at least two -vertices by Claim 7. Thus, by (R2) and (R3).
Assume that z is adjacent to exactly one 3-vertex. Then, it is adjacent to at most one 4-vertex by Claim 7. If z is not adjacent to any 4-vertex, then it is adjacent to two -vertices and one -vertex by Claim 7. Thus, by (R1)∼(R3). If z is adjacent to exactly one 4-vertex, then it is adjacent to two -vertices by Claim 7. Thus, by (R1)∼(R3).
Assume that z is adjacent to two 3-vertices. Then, z is adjacent to two -vertices. If z is adjacent to two 6-vertices, then by (R1)∼(R3) as each 6-vertex is not adjacent to any 3-vertex in by Claim 7. If it is adjacent to one -vertex and one -vertex, then by (R1)∼(R3).
Suppose that the configuration of is in Figure 2. Then, z is incident with two false 3-faces.
Assume that z is not adjacent to any 3-vertex. Then z is adjacent to at most two 4-vertices by Claim 7.
Let z be not adjacent to any 4-vertex. Then, it is adjacent to four -vertices by Claim 7. Thus, by (R2) and (R3).
Let z be adjacent to exactly one 4-vertex. Then, it is adjacent to one -vertices and two -vertices by Claim 7. Without loss of generosity, suppose that . Then, . If , then is not adjacent to any 3-vertex in H (and thus ) by Claim 7 as it is adjacent to one 4-vertex in H. Thus, by (R2) and (R3). If , then by (R2) and (R3).
Let z be adjacent to two 4-vertices. As two 4-vertices are not adjacent in H (and thus ) by Claim 7, without loss of generosity, suppose that . Then, by Claim 7. Note that if , then is not adjacent to any 3-vertex in H (and thus ) by Claim 7 as it is adjacent to one 4-vertex in H. If , then by (R2) and (R3). If and , then by (R2) and (R3). If and , then by (R2) and (R3). If and , then by (R2) and (R3).
Assume that z is adjacent to exactly one 3-vertex. Without loss of generosity, suppose that . Then by Claim 7.
Let . As is adjacent to a 3-vertex in H, it is not adjacent to any other -vertex except in H by Claim 7. Thus, and . Therefore, by (R1)∼(R3).
Let . If , then it is not adjacent to any other -vertex except in H as is adjacent to a 3-vertex in H by Claim 7. Thus, . Therefore, by (R1)∼(R3). If , then by (R1)∼(R3).
Assume that z is adjacent to two 3-vertices. Then, z is adjacent to two -vertices by Claim 7. Thus, by (R1)∼(R3).
Suppose that the configuration of is in Figure 2. Then, z is incident with two false 3-faces. Note that is also incident with two false 3-faces.
Assume that z is not adjacent to any 3-vertex. Then, z is adjacent to at most two 4-vertices by Claim 7.
Let z be not adjacent to any 4-vertex. Then, it is adjacent to four -vertices by Claim 7. Thus, by (R2) and (R3).
Let z be adjacent to exactly one 4-vertex.
Let . Then, by Claim 7. If , then is not adjacent to any 3-vertex in H (and thus ) by Claim 7 as it is adjacent to one 4-vertex in H. Note that . Thus, by (R2) and (R3). If , by (R2) and (R3).
By symmetry, we know that when .
Let . Then, by Claim 7. Thus, by (R2) and (R3).
Let . Then by Claim 7. Note that . If , then is not adjacent to any 3-vertex in H (and thus ) by Claim 7 as it is adjacent to one 4-vertex in H. Thus, by (R2) and (R3). If , by (R2) and (R3).
Let z be adjacent to two 4-vertices. As two 4-vertices are not adjacent in H by Claim 7, . Then, by Claim 7. Without loss of generosity, suppose that . Then, by Claim 7. Note that if , then is not adjacent to any 3-vertex in H (and thus ) by Claim 7 as it is adjacent to one 4-vertex in H. If , then by (R2) and (R3). If and , then by (R2) and (R3). If and , then by (R2) and (R3). If and , then by (R2) and (R3).
Assume that z is adjacent to exactly one 3-vertex.
Let . Then, by Claim 7.
Suppose that . As is adjacent to a 3-vertex in H, it is not adjacent to any other -vertex except in H by Claim 7. Thus, . As and , by (R1)∼(R3).
Suppose that . As and , Thus, by (R1)∼(R3)
By symmetry, we know that when .
Let . Then, by Claim 7. If , then is not adjacent to any other 3-vertex in H (and thus ) by Claim 7 as it is adjacent to a 3-vertex in H. Thus, by (R1)∼(R3). If , then by (R1)∼(R3).
Let . Then, by Claim 7.
Suppose that . Since is adjacent to a 3-vertex in H, it is not adjacent to any other -vertex in H by Claim 7. Thus, by Claim 7. Thus, by (R1)∼(R3).
Suppose that . Note that . If , then by Claim 7. Thus, by (R1)∼(R3) since . If , then by Claim 7. Thus, by (R1)∼(R3) as .
Assume that z is adjacent to two 3-vertices. Then, as two 3-vertices are not adjacent in H (and thus ) by Claim 7. Thus, . For and , without loss of generosity, suppose that . Then, . If , then is not adjacent to any 3-vertex in as it is adjacent to one 3-vertex in H. Thus, by (R1)∼(R3) as . If , by (R1)∼(R3) as .
Therefore, for each false 4-vertex .
In summary, for each , which contradicts . The proof of Theorem 3 is completed.
Funding
This research was funded by Shangluo Science and Technology Plan Project (No. SK2017-40).
Acknowledgments
The author is grateful to all anonymous reviewers and the editor for their inspiring and constructive comments and suggestions.
Conflicts of Interest
The author declares no conflict of interest.
Appendix A
Note that in Claim 2 and P in the following have the same set of the monomials with highest degree and nonzero coefficient in their expansions.
The NoteBook of Mathematica to compute the coefficient.
% INPUT
% Claim 2
P = (x − y)(x − x)(x − y)(x − x)(x + y)(x + y)
Cp = Coefficient[P, xyx]
% To calculate the coefficient of xyx
% OUTPUT
Cp = 5
References
- Bondy, J.A.; Murty, U.S.R. Graph Theory. In GTM; Springer: Berlin/Heidelberg, Germany, 2008; Volume 244. [Google Scholar]
- Pilśniak, M.; Woźniak, M. On the total-neighbor-distinguishing index by sums. Graphs Comb. 2015, 31, 771–782. [Google Scholar] [CrossRef]
- Yao, J.; Yu, X.; Wang, G.; Xu, C. Neighbor sum distinguishing total coloring of 2-degenerate graphs. J. Comb. Optim. 2017, 34, 64–70. [Google Scholar] [CrossRef]
- Li, H.; Liu, B.; Wang, G. Neighbor sum distinguishing total coloring of K4-minor-free graphs. Front. Math. China 2013, 8, 1351–1366. [Google Scholar] [CrossRef]
- Yang, D.; Sun, L.; Yu, X.; Wu, J.; Zhou, S. Neighbor sum distinguishing total chromatic number of planar graphs with maximum degree 10. Appl. Math. Comput. 2017, 314, 456–468. [Google Scholar] [CrossRef]
- Alberson, M. Chromatic number, independent ratio, and crossing number. Ars Math. Contemp. 2008, 1, 1–6. [Google Scholar] [CrossRef]
- Song, C.; Xu, C. Neighbor sum distinguishing total colorings of IC-planar graphs with maximum degree 13. J. Comb. Optim. 2020, 39, 293–303. [Google Scholar] [CrossRef]
- Song, C.; Jin, X.; Xu, C. Neighbor sum distinguishing total coloring of IC-planar graphs with short cycle restrictions. Discret. Appl. Math. 2020, 279, 202–209. [Google Scholar] [CrossRef]
- Song, W.; Duan, Y.; Miao, L. Neighbor sum distinguishing total coloring of triangle free IC-planar graphs. Acta Math. Sin. (Engl. Ser.) 2020, 36, 292–304. [Google Scholar] [CrossRef]
- Qu, C.; Wang, G.; Yan, G.; Yu, X. Neighbor sum distinguishing total choosability of planar graphs. J. Comb. Optim. 2016, 32, 906–916. [Google Scholar] [CrossRef]
- Wang, J.; Cai, J.; Qiu, B. Neighbor sum distinguishing total choosability of planar graphs without adjacent triangles. Theoret. Comput. Sci. 2017, 661, 1–7. [Google Scholar] [CrossRef]
- Song, W.; Miao, L.; Duan, Y. Neighbor sum distinguishing total choosability of IC-planar graphs. Discuss. Math. Graph Theory 2020. [Google Scholar] [CrossRef]
- Alon, N. Combinatorial nullstellensatz. Combin. Probab. Comput. 1999, 8, 7–29. [Google Scholar] [CrossRef]
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |
© 2021 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).