Discrete Mathematics 340 (2017) 119–123

Contents lists available at ScienceDirect

Discrete Mathematics
journal homepage: www.elsevier.com/locate/disc

Adjacent vertex distinguishing total coloring of graphs with

maximum degree 4
You Lu a , Jiaao Li b , Rong Luo b , Zhengke Miao c, *
a
Department of Applied Mathematics, School of Science, Northwestern Polytechnical University, Xi’an, Shaanxi, 710072, China
b
Department of Mathematics, West Virginia University, Morgantown, WV 26505, United States
c
School of Mathematics and Statistics, Jiangsu Normal University, Xuzhou, Jiangsu, 221116, China

article info a b s t r a c t
Article history: Let k be a positive integer. An adjacent vertex distinguishing (for short, AVD) total
Received 17 April 2015 k-coloring φ of a graph G is a proper total k-coloring of G such that no pair of adjacent
Received in revised form 4 January 2016 vertices have the same set of colors, where the set of colors at a vertex v is {φ (v )} ∪
Accepted 19 July 2016
{φ (e) : e is incident to v}. Zhang et al. conjectured in 2005 that every graph with maximum
degree ∆ has an AVD total (∆ + 3)-coloring. Recently, Papaioannou and Raftopoulou
Keywords: confirmed the conjecture for 4-regular graphs. In this paper, by applying the Combinatorial
Adjacent vertex distinguishing total Nullstellensatz, we verify the conjecture for all graphs with maximum degree 4.
coloring © 2016 Elsevier B.V. All rights reserved.
Combinatorial Nullstellensatz
Maximum degree

1. Introduction

All graphs considered in this paper are simple and undirected. We follow the standard notation and terminology as can
be found in [6]. Let G = (V (G), E(G)) be a graph and T (G) = V (G) ∪ E(G). For a vertex x ∈ V (G), we use NG (v ) and EG (v ) to
denote the set of vertices adjacent to v and the set of edges incident to v , respectively. An ℓ-vertex or ℓ− -vertex of G is a
vertex of degree ℓ or at most ℓ, respectively. Let Vℓ (G) and Vℓ− (G) be the sets of ℓ-vertices and ℓ− -vertices, respectively, in
G. We also use Vℓ and Vℓ− for short if the graph G is understood in context. The maximum degree of G is denoted by ∆(G).
Let k be a positive integer and [k] = {1, 2, . . . , k}. A mapping φ : T (G) → [k] is a proper total k-coloring if, for any two
adjacent∑ or incident elements z1 , z2 ∈ T (G), it is φ (z1 ) ̸ = φ (z2 ). Let Cφ (v ) = {φ (v )} ∪ {φ (e) : e ∈ EG (v )} and mφ (v ) =
φ (v ) + e∈EG (v) φ (e) for any vertex v ∈ V (G). A proper total k-coloring φ of G is adjacent vertex distinguishing (for short,
AVD) if Cφ (u) ̸ = Cφ (v ) whenever uv ∈ E(G). The AVD total chromatic number χat (G) is the smallest integer k such that G has
an AVD total k-coloring.
The AVD total coloring is related to vertex-distinguishing edge coloring which requires that every pair of vertices receives
different the sets of colors. The vertex-distinguishing edge coloring was introduced by Burris and Schelp [7], and indepen-
dently by Černý et al. [9] (under the notion of observability). This type of coloring has been well studied over the last decade
(see, for example, [2–5]). It was later extended to require only adjacent vertices to be distinguished by Zhang et al. [14],
which was in turn extended to total coloring [13].
Zhang et al. [13] determined χat (G) for some basic graphs such as complete graphs and complete bipartite graphs and
made the following conjecture.

Conjecture 1.1 ([13]). For any graph G, χat (G) ≤ ∆(G) + 3.

* Corresponding author.
E-mail addresses: luyou@nwpu.edu.cn (Y. Lu), joli@mix.wvu.edu (J. Li), rluo@math.wvu.edu (R. Luo), zkmiao@jsnu.edu.cn (Z. Miao).

http://dx.doi.org/10.1016/j.disc.2016.07.011
0012-365X/© 2016 Elsevier B.V. All rights reserved.
120 Y. Lu et al. / Discrete Mathematics 340 (2017) 119–123

Chen [8] and Wang [12], independently, confirmed Conjecture 1.1 for graphs with maximum degree 3. Later, Hulgan [10]
presented a concise proof on this result. Recently Papaioannou and Raftopoulou [11] verified Conjecture 1.1 for 4-regular
graphs.

Theorem 1.2 ([11]). For any 4-regular graph G, χat (G) ≤ 7.

The aim of this paper is to extend Theorem 1.2 from 4-regular graphs to graphs with maximum degree 4. We prove the
following result.

Theorem 1.3. For any graph G with maximum degree 4, χat (G) ≤ 7.
We use a polynomial method based on the Combinatorial Nullstellensatz due to Alon [1]. In fact, in Section 3, we prove
a stronger result as follows.

Theorem 1.4. Every graph with maximum degree 4 has a proper total 7-coloring satisfying:
(i) For any two adjacent 4-vertices u and v , Cφ (u) ̸ = Cφ (v );
(ii) For any two adjacent 3− -vertices u and v , mφ (u) ̸ = mφ (v ).

Remark. If mφ (u) ̸ = mφ (v ), then the sets of colors must be different. Also, in the definition of AVD total coloring it requires
that any two adjacent vertices have different color sets. Theorem 1.4 does not cover a 4-vertex that is adjacent to a 3− -vertex.
Of course, in a proper total coloring, two adjacent vertices of different degrees also have different color sets.

2. A polynomial associated with AVD total coloring

Let G be a graph with maximum degree 4 and H be an induced subgraph in G[V3− ]. An H-partial AVD total 7-coloring of G
is a mapping φ : T (G) − T (H) → [7], satisfying the following two conditions:

(a) For any two adjacent or incident elements z1 , z2 ∈ T (G) − T (H), φ (z1 ) ̸ = φ (z2 );
(b) For uv ∈ E(G − H), Cφ (u) ̸ = Cφ (v ) if dG (u) = dG (v ) = 4, and mφ (u) ̸ = mφ (v ) if dG (u) ≤ 3 and dG (v ) ≤ 3.

H is called reducible if every H-partial AVD total 7-coloring can be extended to a proper total coloring of G satisfying the
conditions of Theorem 1.4. We will use a polynomial method to prove Theorem 1.4. For this, we need the following theorem,
known as the Combinatorial Nullstellensatz due to Alon [1].

Theorem 2.1 ([1]). Let F be an arbitrary field and P ∈ F[x1 , . . . , xn ] with degree deg(P) =
∑n
j=1 ij , where each ij is a nonnegative
i
integer. If the coefficient of the monomial . . . in P is nonzero, and if S1 , . . . , Sn are subsets of F with |Sj | > ij , then there are
x11 xinn
s1 ∈ S1 , . . . , sn ∈ Sn such that P(s1 , . . . , sn ) ̸ = 0.
Let H be an induced subgraph of G[V3− ]. Denote V (H) = {v1 , . . . , vh } and E(H) = {e1 , . . . , ek }. Each element z ∈
T (H) = V (H) ∪ E(H) is associated with a variable xz . Let D be an arbitrary orientation of H and φ be an H-partial AVD total
7-coloring of G.∑For each vertex v ∈ V (H), ND+ (v ) is the set of arcs with v as the initial vertex. For each vertex v ∈ V (H), let
µH (v ) = xv + e∈EH (v) xe and
⎛ ⎞
∏ ∏
PD,φ (H ; v ) = ⎝(xv − φ (u))(xv − φ (uv )) (xe − φ (uv ))⎠
u∈NG (v )\V (H) e∈EH (v )
⎛ ⎞
∏ ∑ ∑
· ⎝ µH ( v ) + φ (e) − φ (u) − φ (e)⎠
u∈(V3− ∩NG (v ))\V (H) e∈EG (v )\EH (v ) e∈EG (u)
⎛ ⎞
∏ ∑ ∑
· (xv − xu ) ⎝µH (v ) + φ (e) − µH (u) − φ (e)⎠
+
u∈ND (v ) e∈EG (v )\EH (v ) e∈EG (u)\EH (u)
∏ ∏
· (xv − xe ) (xei − xej ).
e∈EH (v ) ei ,ej ∈EH (v )
i<j

Remark. In PD,φ (H ; v ), the first product assures that v and every edge e ∈ EH (v ) would have different colors than its incident
elements in T (G) − T (H); while the last two products (together with some parts of the third product) assure that v and every
edge e ∈ EH (v ) would have different color than its incident elements in T (H). Moreover, the second and third products
guarantee mφ (u) ̸ = mφ (v ) for any vertex u ∈ V3− ∩ NG (v ).
 Y. Lu et al. / Discrete Mathematics 340 (2017) 119–123 121

By the Combinatorial Nullstellensatz (Theorem 2.1), we need to find the existence of certain monomials of degree deg(P)
with nonzero coefficient in the expansion of PD,φ (H ; v ). Since φ (x) is a constant for each element x ∈ T (G) − T (H), one can
drop each constant term from PD,φ (H ; v ) to only consider its homogeneous part which can be expressed as follows.

⎛ ⎞|NG (v)\NH (v)|

(µH (v ))|(V3− ∩NG (v))\V (H)|

∏
P̃D (H ; v ) = ⎝x2v xe ⎠
e∈EH (v )
∏ ∏ ∏
· (xv − xu )(µH (v ) − µH (u)) (xv − xe ) (xei − xej ).
+
u∈ND (v ) e∈EH (v ) ei ,ej ∈EH (v )
i<j

Note that v∈V (H) PD,φ (H ; v ) and v∈V (H) P̃D (H ; v )

∏ ∏
are independent of the orientation D. Thus we define the following two
polynomials.
∏
Pφ (H) = Pφ (xv1 , xv2 , . . . , xvh , xe1 , xe2 , . . . , xek ) = PD,φ (H ; v ),
v∈V (H)
∏
P̃ (H) = P̃ (xv1 , xv2 , . . . , xvh , xe1 , xe2 , . . . , xek ) = P̃D (H ; v ).
v∈V (H)

Lemma 2.2. Let G be a graph with maximum degree 4 and let ∏ H be an induced subgraph of G[V3− ] such that there exists an
jz
H-partial AVD total 7-coloring
∑ of G. If P̃ (H) has a monomial z ∈T (H) xz with nonzero coefficient such that jz ≤ 6 for each
z ∈ T (H) and deg(P̃ (H)) = z ∈T (H) jz , then H is reducible.

∏ jz ∑
Proof. Let z ∈T (H) xz be a monomial of P̃ (H) with nonzero coefficient such that deg(P̃ (H)) = z ∈T (H) jz and jz ≤ 6 for
j
∑ z ∈ T (H). Let φ be an H-partial AVD total 7-coloring of G. Then z ∈T (H) xz is also a monomial of Pφ (H) such that
z
∏
each
z ∈T (H) jz = deg(Pφ (H)) and its coefficient in Pφ (H) is the same as its coefficient in P̃ (H). Thus by Theorem 2.1, for each
z ∈ T (H), there is an integer az ∈ [7] such that
Pφ (av1 , av2 , . . . , avh , ae1 , ae2 , . . . , aek ) ̸ = 0.

Hence φ can be extended to a coloring of G satisfying the conditions of Theorem 1.4 by coloring each z ∈ T (H) with az . By
definition, H is reducible. ■
The following lemma is needed in the proof of the main result and its proof is straightforward and thus omitted.

Lemma 2.3. Let F be an arbitrary field and P , Q , R ∑

be three polynomials in F[x1 , . . . , xn ] with P = Q · R. If the coefficient of the
i
monomial x11 . . . xinn in P is nonzero and deg(P) =
n ′
j=1 ij , then there are nonnegative integers ij ≤ ij for each j ∈ [n] such that
i′ i′
and the coefficient of the monomial x11 . . . xnn in Q is also nonzero.
∑n ′
deg(Q ) = j =1 i j

3. Graphs with ∆ = 4

In this section, we will prove Theorem 1.4. Let us start with the special case of graphs in which V3− is an independent set
of vertices. Note that if V3− is an independent set of vertices, then any AVD total 7-coloring of G satisfies the two conditions
of Theorem 1.4.

Lemma 3.1. Let G be a connected graph with maximum degree 4. If V3− is an independent set of vertices in G, then G has an AVD
total 7-coloring.

Proof. For i ∈ {1, 2, 3, 4}, let Gi be a copy of G, and use vi ∈ V (Gi ) to denote the copy of v ∈ V (G) in Gi . Let G′ = ∪4i=1 Gi .
Then we augment G′ to a new graph G′′ by adding some new edges to G′ such that for each v ∈ V (G), the induced subgraph
G′′ [{v1 , v2 , v3 , v4 }] is (4 − dG (v ))-regular in G′′ . It is obvious that G′′ is a 4-regular graph. By Theorem 1.2, G′′ has an AVD total
7-coloring ψ . Since V3− is an independent set of G, ψ also induces an AVD total 7-coloring of G1 (and thus of G) satisfying
the two conditions of Theorem 1.4. ■
Outline of the proof of Theorem 1.4. The proof is based on a minimal counterexample. By the minimality of G, we first
show that no connected induced subgraph in G[V3− ] with at least two vertices is reducible. Applying this claim repeatedly,
we show that G[V3− ] contains certain configurations. Finally we prove that such configurations are indeed reducible, which
contradicts to the first claim.

Proof of Theorem 1.4. Suppose that Theorem 1.4 does not hold. Let G be a counterexample to Theorem 1.4 such that |E(G)|
is minimum. Obviously, G is connected and V3− is not an independent set of G by Lemma 3.1.
122 Y. Lu et al. / Discrete Mathematics 340 (2017) 119–123

Fig. 1. H and the corresponding variables in Claim 3.3.

Claim 3.1. Let H be a connected subgraph induced by at least two vertices of V3− . Then G has an H-partial AVD total 7-coloring
and thus H is not reducible.

Proof. Since H is connected and has at least two vertices, it is |E(H)| ≥ 1. By the minimality of G, G − E(H) has a proper total
7-coloring ψ satisfying
(i) For any two adjacent 4-vertices u and v , Cψ (u) ̸ = Cψ (v );
(ii) For any two adjacent 3− -vertices u and v , mψ (u) ̸ = mψ (v ).
Uncoloring every vertex in V (H), we obtain a proper total 7-coloring φ on T (G) − T (H). It is easy to see that φ is an H-partial
AVD total 7-coloring of G. Since G is a counterexample, φ cannot be extended to an AVD total 7-coloring of G satisfying the
conditions of Theorem 1.4 and thus H is not reducible. ■
The following claim is a direct corollary of Claim 3.1 and Lemma 2.2.

iz
∏
Claim 3.2. Let H be ∑ an induced subgraph of G[V3− ] with |E(H)| ̸ = 0. Then P̃ (H) has no monomials z ∈T (H) xz with nonzero
coefficient such that z ∈T (H) iz = deg(P̃ (H)) and iz ≤ 6 for each z ∈ T (H).

Claim 3.3. For any edge u1 u2 in G[V3− ], we have the following two assertions.
(1) dG (u1 ) = dG (u2 ) = 3 and thus V2 is an independent set.
(2) At least one of u1 and u2 is adjacent to three 3-vertices.

Proof. For i = 1, 2, let di = dG (ui ) − 1 and ki = |NG (ui ) ∩ V4 |. Since u1 u2 ∈ E(G[V3− ]), we have 0 ≤ ki ≤ di ≤ 2. Let H be the
induced subgraph of G with vertex set V (H) = {u1 , u2 } and edge set {u1 u2 }, i.e. H = u1 u2 . For convenience denote xi = xui
for i = 1, 2 and y = xu1 u2 . Then by the definition of P̃ (H), we have

P̃ (H) = (x21 y)d1 (x1 + y)d1 −k1 (x1 − x2 )(x1 + y − x2 − y)(x1 − y) · (x22 y)d2 (x2 + y)d2 −k2 (x2 − y).

(1) We first prove that dG (u1 ) = dG (u2 ) = 3. Suppose to the contrary that, without loss of generality, dG (u1 ) ≤ 2 and
dG (u2 ) ≤ 3. Then 0 ≤ d1 ≤ 1 and 0 ≤ d2 ≤ 2. It is easy to check that deg(P̃ (H)) = 4d1 + 4d2 − k1 − k2 + 4 and
3d −k +3 2d
that the coefficient of the monomial x1 1 1 x2 2 yd1 +2d2 −k2 +1 of P̃ (H) is −1. We have that 3d1 − k1 + 3 ≤ 6, 2d2 ≤ 4,
d1 + 2d2 − k2 + 1 ≤ 6, and 3d1 − k1 + 3 + 2d2 + d1 + 2d2 − k2 + 1 = deg(P̃ (H)), a contradiction to Claim 3.2. This proves
dG (u1 ) = dG (u2 ) = 3.
Note that the above assertion implies that V2− is an independent set of G[V3− ]. Otherwise, we would have an edge u1 u2
in G[V3− ] with min{dG (u1 ), dG (u2 )} ≤ 2, a contradiction to the fact that dG (u1 ) = dG (u2 ) = 3.
(2) We now prove that at least one of u1 and u2 is adjacent to three 3-vertices (See Fig. 1). Suppose to the contrary that
both u and v are adjacent to at most two 3-vertices. Then k1 ≥ 1 and k2 ≥ 1 since dG (u) = dG (v ) = 3 and V2− is an
independent set of G[V3− ] by the first part of the claim. Let P = P̃ (H) · (x1 + y)k1 −1 (x2 + y)k2 −1 . Since d1 = d2 = 2,
P = P̃ (H) · (x1 + y)k1 −1 (x2 + y)k2 −1
= (x21 y)2 (x22 y)2 (x1 + y)(x2 + y)(x1 − y)(x2 − y)(x1 − x2 )2
= −2x61 x62 y6 − x41 x42 y6 (x41 − 2x31 x2 − 2x1 x32 + x42 ) + x41 x42 y4 (x21 x22 + y4 )(x1 − x2 )2 .
i i
Since deg(P) = 18 and the monomial x61 x62 y6 of P has coefficient −2, by Lemma 2.3, P̃ (H) has a monomial x11 x22 yi3 with
nonzero coefficient such that i1 + i2 + i3 = deg(P̃ (H)) and ij ≤ 6 for j = 1, 2, 3, a contradiction to Claim 3.2. ■

Claim 3.4. If u1 u2 u3 u1 is a triangle in G[V3 ], then each ui is adjacent to three 3-vertices.

Proof. By Claim 3.3, at most one vertex in the triangle is adjacent to a 4-vertex. Suppose to the contrary (without loss of
generality) that u1 is adjacent to one 4-vertex, while both u2 and u3 are adjacent to three 3-vertices. Let H be the triangle
 Y. Lu et al. / Discrete Mathematics 340 (2017) 119–123 123

Fig. 2. H and the corresponding variables in Claim 3.4.

u1 u2 u3 u1 . For convenience denote xi = xui for i = 1, 2, 3, y1 = xu1 u2 , y2 = xu2 u3 , and y3 = xu3 u1 (See Fig. 2). Then we have

P̃ (H) = (x21 y1 y3 )(x1 − y1 )(x1 − y3 )(y1 − y3 )(x1 − x2 )(x1 + y3 − x2 − y2 )

· (x22 y1 y2 )(x2 + y1 + y2 )(x2 − y1 )(x2 − y2 )(y1 − y2 )(x2 − x3 )(x2 + y1 − x3 − y3 )
· (x23 y2 y3 )(x3 + y2 + y3 )(x3 − y2 )(x3 − y3 )(y2 − y3 )(x3 − x1 )(x3 + y2 − x1 − y1 ).
One can check that the coefficient of the monomial x51 x52 x53 y61 y62 y23 in P̃ (H) is −1 and deg(P̃ (H)) = 29, a contradiction to
Claim 3.2. ■
The final step. Let G1 be a connected component of G[V3 ] with |V (G1 )| maximum, and let u1 ∈ V (G1 ) with dG1 (u1 )
minimum. Then |V (G1 )| ≥ 2 since V3 is not independent in G by Lemma 3.1 and V2− is an independent set of G[V3− ] by
Claim 3.3(1). Moreover, dG1 (u1 ) ≤ 2 since ∆(G) = 4 and G is connected. Pick an edge u1 u2 ∈ E(G1 ). By Claim 3.3, u2 is
adjacent to three 3-vertices, and so let u3 ∈ NG1 (u2 ) \ {u1 } ⊆ V3 .
Let H = G[{u1 , u2 , u3 }], k1 = |NG (u1 ) ∩ V4 | and k3 = |NG (u3 ) ∩ V4 |. Then 1 ≤ k1 ≤ 2 and 0 ≤ k3 ≤ 2. By Claim 3.4, H can
not be a triangle and thus H is an induced path u1 u2 u3 in G1 . For convenience denote xi = xui for i = 1, 2, 3, y1 = xu1 u2 , and
y2 = xu2 u3 . Then we have

P̃ (H) = (x21 y1 )2 (x1 + y1 )2−k1 (x1 − y1 )(x1 − x2 )(x1 − x2 − y2 )

· (x22 y1 y2 )(x2 + y1 + y2 )(x2 − y1 )(x2 − y2 )(y1 − y2 )(x2 − x3 )(x2 + y1 − x3 )
· (x23 y2 )2 (x3 + y2 )2−k3 (x3 − y2 ).
Let P = P̃ (H) · (x1 + y1 )k1 −1 (x3 + y2 )k3 . One can check that deg(P) = 29 and the coefficient of the monomial x61 x62 x63 y51 y62 in P is
i i i i i
3. Therefore, by Lemma 2.3, P̃ (H) has a monomial x11 x22 x33 y14 y25 with nonzero coefficient such that ij ≤ 6 for each j = 1, . . . , 5
and i1 + · · · + i5 = deg(P̃ (H)), a contradiction to Claim 3.2. This contradiction completes the proof of Theorem 1.4.

Acknowledgments

Many thanks to the anonymous referees for their careful comments that improved the presentation of this paper. The
first author was partially supported by NSFC under grant number 11201374 and the fourth author was partially supported
by NSFC under grant numbers 11171288 and 11571149.

References

[1] N. Alon, Combinatorial Nullstellensatz, Combin. Probab. Comput. 8 (1999) 7–29.

[2] P.N. Balister, Vertex-distinguishing edge colorings of random graphs, Random Struct. Algorithms 20 (2001) 89–97.
[3] P.N. Balister, A. Kostochka, H. Li, R.H. Schelp, Balanced edge colorings, J. Combin. Theory Ser. B 90 (2004) 3–20.
[4] P.N. Balister, O.M. Riordan, R.H. Schelp, Vertex-distinguishing edge colorings of graphs, J. Graph Theory 42 (2003) 95–109.
[5] C. Bazgan, A. Harkat-Benhamdine, H. Li, M. Woźniak, On the vertex-distinguishing proper edge-coloring of graphs, J. Combin. Theory Ser. B 75 (2)
(1999) 288–301.
[6] J.A. Bondy, U.S.R. Murty, Graph Theory, in: GTM, vol. 244, Springer, 2008.
[7] A. Burris, R. Schelp, Vertex-distinguishing proper edge-colorings, J. Graph Theory 26 (2) (1997) 73–82.
[8] X. Chen, On the adjacent vertex distinguishing total coloring numbers of graphs with ∆ = 3, Discrete Math. 308 (17) (2008) 4003–4007.
[9] J. Černý, M. Horňák, R. Soták, Observability of a graph, Math. Slovaca 46 (1996) 21–31.
[10] J. Hulgan, Concise proofs for adjacent vertex-distinguishing total colorings, Discrete Math. 309 (2009) 2548–2550.
[11] A. Papaioannou, C. Raftopoulou, On the AVDTC of 4-regular graphs, Discrete Math. 330 (2014) 20–40.
[12] H. Wang, On the adjacent vertex distinguishing total chromatic number of the graphs with ∆ = 3, J. Comb. Optim. 14 (2007) 87–109.
[13] Z. Zhang, X. Chen, J. Li, B. Yao, X. Lu, J. Wang, On adjacent-vertex-distinguishing total coloring of graphs, Sci. China Ser. A 48 (2005) 28–299.
[14] Z. Zhang, L. Liu, J. Wang, Adjacent strong edge coloring of graphs, Appl. Math. Lett. 15 (5) (2002) 623–626.