Volume 2020, Issue 1 6481092
Research Article
Open Access

Some Properties of Double Roman Domination

Hong Yang

Corresponding Author

Hong Yang

School of Information Science and Engineering, Chengdu University, Chengdu 610106, China cdu.edu.cn

Search for more papers by this author
Xiaoqing Zhou

Xiaoqing Zhou

School of Information Science and Engineering, Chengdu University, Chengdu 610106, China cdu.edu.cn

Search for more papers by this author
First published: 14 August 2020
Citations: 4
Academic Editor: Juan L. G. Guirao

Abstract

A double Roman dominating function on a graph G is a function f : V(G)⟶{0,1,2,3} satisfying the conditions that every vertex u for which f(u) = 0 is adjacent to at least one vertex v for which f(v) = 3 or two vertices v1 and v2 for which f(v1) = f(v2) = 2 and every vertex u for which f(u) = 1 is adjacent to at least one vertex v for which f(v) ≥ 2. The weight of a double Roman dominating function f is the value f(V) = ∑uVf(u). The minimum weight of a double Roman dominating function on a graph G is called the double Roman domination number γdR(G) of G. A graph with γdR(G) = 3γ(G) is called a double Roman graph. In this paper, we study properties of double Roman domination in graphs. Moreover, we find a class of double Roman graphs and give characterizations of trees with γdR(T) = γR(T) + k for k = 1,2.

1. Introduction

In this paper, we shall only consider graphs without multiple edges or loops. Let G be a graph, vV(G), and the neighborhood of v in G is denoted by N(v). That is to say, N(v) = {u|uvE(G), uV(G)}. The closed neighborhood N[v] of v in G is defined as N[v] = {v} ∪ N(v). The complementary graph of G is denoted by . A vertex of degree one is called a leaf. A graph is trivial if it has a single vertex. The degree of a vertex v is denoted by d(v), i.e., d(v) = |N(v)|. Denote by Kn, Pn, and Cn the complete graph, path, and cycle on n vertices, respectively. The maximum degree and the minimum degree of a graph are denoted by Δ(G) and δ(G), respectively. For a set SV(G), the graph induced by S is denoted by G[S]. Let eE(G), and we denote by G/e the graph obtained from G by contracting the edge e. For an edge eE(G), we denote by Ge the graph obtained from G by deleting e.

A subset D of the vertex set of a graph G is a dominating set if every vertex not in D has at least one neighbour in D. The domination number γ(G) is the minimum cardinality of a dominating set of G.

The domination and its variations of graphs have attracted considerable attention [1, 2]. Many varieties of dominating sets are listed in the book Fundamentals of Domination in Graphs [3]. However, Roman domination and double Roman domination are not listed in this book. Roman domination and double Roman domination appear to be a new variety of interest [411].

A Roman dominating function (RDF) of a graph G is a function f : V(G)⟶{0,1,2} such that every vertex u for which f(u) = 0 is adjacent to at least one vertex v for which f(v) = 2. The weight w(f) of a Roman dominating function f is the value w(f) = ∑uV(G)f(u). The minimum weight of a Roman dominating function on a graph G is called the Roman domination number γR(G) of G. An RDF f of G with w(f) = γR(G) is called a γR(G) function.

A double Roman dominating function (DRDF) on a graph G is a function f : V(G)⟶{0,1,2,3} satisfying the condition that every vertex u for which f(u) = 0 is adjacent to at least one vertex v for which f(v) = 3 or two vertices v1 and v2 for which f(v1) = f(v2) = 2 and every vertex u for which f(u) = 1 is adjacent to at least one vertex v for which f(v) ≥ 2. The weight w(f) of a double Roman dominating function f is the value w(f) = ∑uV(G)f(u). The minimum weight of a double Roman dominating function on a graph G is called the double Roman domination number γdR(G) of G. A DRDF f of G with w(f) = γdR(G) is called a γdR(G) function. We denote by wS(f) the weight of a double Roman dominating function f in SV(G), i.e., wS(f) = ∑xSf(x).

Beeler et al. [12] initiated the study of the double Roman domination in graphs. They showed that 2γ(G) ≤ γdR(G) ≤ 3γ(G) and defined a graph G to be double Roman if γdR(G) = 3γ(G). Moreover, they suggest to find double Roman graphs.

In this paper, we study properties of double Roman domination in graphs and show that the double Roman domination problem is NP-complete for bipartite graphs. Moreover, we find a class of double Roman graphs and give characterizations of trees with γdR(T) = γR(T) + k for k = 1,2.

2. Properties of Double Roman Domination

Proposition 1 (see [12].)In a double Roman dominating function of weight γdR(G), no vertex needs to be assigned the value 1.

By Proposition 1, when we consider a γdR(G) function, we assume no vertex has been assigned the value 1.

Proposition 2 (see [12].)

  • (i)

    Let G be a graph and f = (V0, V1, V2) be a γR(G) function. Then, γdR(G) ≤ 2|V1| + 3|V2|.

  • (ii)

    For any graph G, γdR(G) ≤ 2γR(G) with equality if and only if .

Proposition 3 (see [12].)

  • (i)

    For every graph G, γR(G) < γdR(G).

  • (ii)

    If f = (V0, ∅, V2, V3) is any γdR(G) function, then γR(G) ≤ 2(|V2| + |V3|) = γdR(G) − |V3|.

The following result is immediate.

Proposition 4. For any graph G, γdR(G) ≥ (3|V(G)|/Δ(G) + 1).

Proof. The desired inequality obviously holds if Δ(G) ≤ 1. In order to prove the proposition for Δ(G) ≥ 2, we introduce the discharging approach. Let f = (V0, ∅, V2, V3) be a γdR(G) function. The initial charge of every vertex vV(G) is set to be s(v) = f(v). We apply the discharging procedure defined by applying the following rule.

For each vertex vV3, we send 3/(d(v) + 1) charge to each adjacent vertex in V0. Then, the final charge of v is satisfying with s(v) = (3/(d(v) + 1)) ≥ (3/(Δ(G) + 1)).

For each vertex vV2, we send (2/(d(v) + 1)) − (3/(d(v)(Δ(G) + 1))) = ((2Δ(G) − 1)/(d(v)(Δ(G) + 1))) ≥ ((2Δ(G) − 1)/(Δ(G)(Δ(G) + 1))) charge to each adjacent vertex in V0. Then, the final charge of v is satisfying with s(v) = (3/(d(v) + 1)) ≥ (3/(Δ(G) + 1)).

For each vertex vV0, by the definition of double Roman domination, v has a neighbor assigned 3 or two vertices u1 and u2 assigned 2. Due to the discharging rule above, if v has a neighbor u assigned 3, v receives charge from u. We have s(v) ≥ (3/(Δ(G) + 1)).

If v has two vertices u1 and u2 assigned 2, v receives charge from u1 and u2. We have s(v) ≥ ((4Δ(G) − 2)/(Δ(G)(Δ(G) + 1))) ≥ (3/(Δ(G) + 1)). Thus, γdR(G) = ∑vV(G)f(v) = ∑vV(G)s(v) ≥ ((3|V(G)|)/(Δ(G) + 1)). The proof is complete.

Proposition 5. Let G be a graph. If γR(G) + 1 = γdR(G) and f = (V0, ∅, V2, V3) is a γdR(G) function,

  • (i)

    then  |V3| ≤ 1

  • (ii)

    if |V3| = 1, then |V2| = 0 and there exists a vertex v with degree |V(G)| − 1.

Proof.

  • (i)

    By Proposition 3, we have γR(G) ≤ γdR(G) − |V3| = γR(G) + 1 − |V3|. So we have |V3| ≤ 1.

  • (ii)

    If |V3| = 1, let V3 = {v} and H = G[V(G) − N[V3]]. We have the following claim.

Claim 1. H is empty.

Proof. Otherwise, any vertex in H assigned with 0 has at least two vertices assigned with 2. Let w be a vertex in H assigned with 2, we consider a function f with f(w) = 1, f(v) = 2, and f(x) = f(x) for any xV(G) − {v, w}. Then, f is an RDF of G with weight γdR(G) − 2. So γR(G) ≤ γdR(G) − 2, a contradiction.

Since H is empty, we have v as a vertex with degree |V(G)| − 1. Now, we have γdR(G) = 3 and γR(G) = 2, and so the result holds.

Theorem 1. For every graph G on n vertices without isolated vertex, γdR(G) ≤ 3n − (3n/(2(1 + δ(G))))e1/δ(G).

Proof. Clearly, we have δ(G) ≥ 1. We select a subset S of V(G), where each vertex is selected with probability p independently. Let T = V(G) − N[S], we consider a function f : V(G)⟶{0,2,3} with f(x) = 3 for xS, f(x) = 2 for xT, and f(x) = 0 for other vertex x. Then, f is a DRDF of G. We have γdR(G) ≤ 3|S| + 2|T|. When we consider the expectation, we also have γdR(G) ≤ E[3|S| + 2|T|] = 3E[|S|] + 2E[|T|]. First, it is clear that E[|S|] = np. For each vertex with degree d(x), if neither x nor any neighbor is selected, then xT. So we have P(xT) = (1 − p)1+d(x), and thus, E[|T|] ≤ n(1 − p)1+δ(G). Consequently, γdR(G) ≤ 3np + 2n(1 − p)1+δ(G). Let F(p) = 3np + 2n(1 − p)1+δ(G). By F(p) = 3np − 2n(1 + δ(G))(1 − p)δ(G) = 0, the maximum of F is given by Fmax = 3n − (3n/(2(1 + δ(G))))e1/δ(G) at p = 1 − (3/(2(1 + δ(G))))e1/δ(G) ∈ (0,1). Thus, γdR(G) ≤ 3n − (3n/(2(1 + δ(G))))e1/δ(G).

If G is a graph with some isolated vertices, then δ(G) = 0. Let W be the set of isolated vertices of G and let G = GW. Therefore, δ(G) ≥ 1. Because all isolated vertices must be assigned 3, it is easy to prove that , where n = |V(G)|.

Proposition 6. If G is a connected graph of order n, then γdR(G) + 1 = 2γR(G) if and only if there exists a vertex v of degree n + 1 − ((γdR(G) + 1)/2) in G.

Proof.

  • (⇒) Let f = (V0, V1, V2) be a γR(G) function with minimum |V1|. Then, we have V1 being independent. Together with Proposition 2, we have γdR(G) ≤ 2|V1| + 3|V2| ≤ 2|V1| + 4|V2| = 2γR(G) = γdR(G) + 1. Then, we have |V2| ≤ 1. If |V2| = 0, we have |V0| = 0, and thus γR(G) = n = |V1|. Since V1 is independent, it is impossible. If |V2| = 1, we have γdR(G) = 2|V1| + 3|V2| < 2|V1| + 4|V2| = 2γR(G) = γdR(G) + 1. Let V2 = {v}, V0 = N(v), and V1 = V(G) − V0V2. Then, we have γR(G) = n − 1 − d(v) + 2 and so d(v) = n + 1 − γR(G) = n + 1 − ((γdR(G) + 1)/2).

  • (⇐) By Proposition 2 (ii), we have γdR(G) + 1 ≤ 2γR(G) for a connected graph G. Assume G contains a vertex v of degree n + 1 − (γdR(G) + 1)/2 in G. Let V2 = {v}, V0 = N(v), and V1 = V(G) − V0V2. Then, f = (V0, V1, V2) is a γR(G) function and so γR(G) ≤ |V1| + 2|V2| = (γdR(G) + 1)/2. Hence, γdR(G) + 1 ≥ 2γR(G).

Let ℱ be the family of connected graphs G such that for any γdR(G) function f = (V0, ∅, V2, V3), we have |V3| = 0.

Proposition 7. Let G ∈ ℱ, then

  • (i)

    G contains no strong support vertex

  • (ii)

    if f(u) = f(v) = 2 for an edge uvE(G), then Ge ∈ ℱ and G/e ∈ ℱ.

Proof.

  • (i)

    Suppose v be a strong support vertex, then there exist two leaves x, yN(v). Since f(v) ≠ 3, we have f(x) = f(y) = 2. Now consider the function f with f(z) = 0 for any zL(v), f(v) = 3, and f(z) = f(z) for any zV(G)∖L[v]. Then, f is a DRDF of G with fewer weight than f, a contradiction.

  • (ii)

    If f(u) = f(v) = 2 for an edge uvE(G), we have f as also a DRDF of Ge. So γdR(Ge) ≤ w(f) = γdr(G). Since for any graph G, we have γdR(Ge) ≥ γdR(G), so γdR(Ge) = γdR(G). Suppose to the contrary that there exists a γdR(Ge) function such that f(v) = 3 for a vertex vV(Ge). Then, f is also a γdR(G) function, contradicting with |V3| = 0. For the graph G/e, the proof is similar.

Lemma 1. Let G be a graph on n ≥ 4 vertices, then γdR(G) = 4 if and only if G contains a complete bipartite graph K2,n−2 as a subgraph and Δ(G) ≤ n − 2.

Proof.

  • (⇒) If γdR(G) = 4, then no vertex is assigned with 3 and thus we have two vertices v and w assigned with 2 and the others 0. Also, each vertex 0 must be adjacent to both v and w. Therefore, G contains a complete bipartite graph K2,n−2 as a subgraph. Since γdR(G) > 3, we have Δ(G) ≤ n − 2.

  • (⇐) If G contains a complete bipartite graph K2,n−2 with partitions X, Y (|X| = 2, |Y| = n − 2) as a subgraph and Δ(G) ≤ n − 2. Then, let f(x) = 2 for any xX and f(x) = 0 for xY. Then, f is a DRDF of G and so γdR(G) ≤ 4. Since G contains no vertex with degree |V(G)| − 1, we have γdR(G) ≥ 4.

Note that Δ(G) ≥ n − 2 if G contains a complete bipartite graph K2,n−2 as a subgraph. Thus, Δ(G) ≤ n − 2 can be replaced with Δ(G) = n − 2 in the lemma.

Theorem 2. Let G be a graph on n ≥ 3 vertices, then . Furthermore, equality holds in the upper bound if G or is Kn.

Proof. If G is a graph on n ≥ 3 vertices, we have γdR(G) ≥ 3, and if γdR(G) = 3, then G has a vertex with degree n − 1. But its complement is neither a star nor a graph G with γdR(G) = 4 (see the graph stated in Lemma 1). So we have and thus, . If G is a star, then and so the lower bound is attainable.

Let v be a vertex with maximum degree Δ(G); consider a function f with f(v) = 3, f(x) = 0 for any xN(v), and f(x) = 2 for xV(G) − N[x]. Then, f is a DRDF of G and so γdR(G) ≤ w(f) = 2n − 2Δ(G) + 1. Since , we have . Therefore, . It can be seen that if , then Δ(G) = δ(G). Hence, G is k-regular for some k. By symmetry, we may assume that k ≤ (n − 1)/2. Then, if , we have γdR(G) = 2n − 2k + 1 and . Let vV(G). If |N(u)∩N(v)| ≤ k − 2 for some uV(G) − N[v], then f = (N(v) ∪ N(u), ∅, V(G) − N[u] − N[v], {u, v}) is a DRDF of G with fewer weight than 2n − 2k + 1, a contradiction. Therefore, each vertex not in N[v] has at least k − 1 neighbors in N(v). Analogously, each vertex in N[v] has at most 2 neighbors outside N[v]. Then, we have (k − 1)(nk − 1) ≤ 2k. Since k ≤ (n − 1)/2, we have k ≤ 2 + 2/(k − 1). If k = 3, then n = 7. This is impossible. If k = 2, then n ∈ {5,6,7}. If G = C5, we have , and so , a contradiction. If G = C6, we have , and so , a contradiction. If G = C7, we have , and so , a contradiction. If k = 1, then G = (n/2)K2, and we have γdR(G) = 3n/2 and , and so , a contradiction. Therefore, we conclude that . If G = Kn, then and thus the upper bound is attainable.

3. Some Double Roman Graphs

The Cartesian product of graphs G and H is the graph GH with vertex set G × H and (x1, x2)(y1, y2) ∈ E(GH) whenever x1y1E(G) and x2 = y2, or x2y2E(H) and x1 = y1. The Cartesian product is commutative and associative, having the trivial graph as a unit (cf. [13]).

Let f be a double Roman dominating function of CmCn, and we write for i ∈ {0,1,2,3}. When no confusion arise, we simply write as Vi. We use f(i, j) to denote the value f(v) for v = (i, j) ∈ V(CmCn). Let xi be the weight of , i.e., .

Theorem 3. Let m, n ≥ 1. Then, the Cartesian product graphs C5mC5n are double Roman.

Proof. The lower bound follows from Proposition 4. Let V(C5mC5n) = {vij : 0 ≤ i ≤ 5m − 1,0 ≤ j ≤ 5n − 1}, V1 = V2 = ∅, V3 = {v(5i)(5j + 2), v(5i + 1)(5j), v(5i + 2)(5j + 3), v(5i + 3)(5j + 1), v(5i + 4)(5j + 4), 0 ≤ im − 1, 0 ≤ jn − 1}, and V0 = N(V3). Then, N[V3] = V(C5mC5n) and so f = (V0, V1, V2, V3) is a DRDF of C5mC5n with weight 15mn and we have γdR(C5mC5n) ≤ 15mn. Since γ(C5mC5n) = 5mn, we have C5mC5n is double Roman.

4. Trees T with γdR(T) = γR(T) + k

Theorem 4. If T is a tree, then γR(T) + 1 = γdR(T) if and only if T is a star K1,s for s ≥ 1.

Proof.

  • (⇐) If T is a star K1,s for s ≥ 1, it is clear that γdR(T) = 3 and γR(T) = 2, and the theorem holds.

  • (⇒) By Proposition 3, we have |V3| ≤ 1. If |V3| = 1, by Proposition 5, we have T is a star K1,s for some s ≥ 1. If |V2| = 0, then each vertex in T is assigned 2 or 0. Since T is a tree, T has a least two leaves v and w and f(v) = f(w) = 2. If v and w are adjacent to the same vertex x, then we can obtain a DRDF of T with fewer weight by changing f(x) to 3 and f(v) and f(w) to 0 and obtaining a contradiction. If v and w are adjacent to different vertices, we consider a function f with f(w) = 1, f(v) = 1, and f(x) = f(x) for any xV(T) − {v, w}. Then, f is an RDF of T with weight γdR(T) − 2. So γR(T) ≤ γdR(T) − 2, a contradiction.

For a positive integer t, a wounded spider is a star K1,t with at most t − 1 of its edges subdivided. In a wounded spider, a vertex of degree t will be called the head vertex, and the vertices at distance two from the head vertex will be the foot vertices.

Theorem 5. If T is a tree, then γR(T) + 2 = γdR(T) if and only if T is a wounded spider with only one foot or T is obtained by adding an edge between two stars K1,s and K1,t for s, t ≥ 2.

Proof.

  • (⇐) If T is a wounded spider with only one foot, it is clear that γdR(T) = 5 and γR(T) = 3, and the theorem holds. If T is a tree obtained by adding an edge between two stars K1,s and K1,t for s, t ≥ 2, then γdR(T) = 6 and γR(T) = 4, and the theorem holds.

  • (⇒) By Proposition 3, we have |V3| ≤ 2. If |V3| = 2, let V3 = {v, w} and H = G[V(G) − N[w] − N[v]]. Similar to the proof of Theorem 5, we have H as empty. Otherwise, there exists a vertex u in H assigned with 2. Now, we change the function values of v, w, u from 3,3,2 to 2,2,1, respectively, and obtain an RDF of T with weight γdR(T) − 3, a contradiction. In this case, T is a tree obtained by adding an edge between two stars K1,s and K1,t for s, t ≥ 2. If |V3| = 1, let V3 = {v} and H = G[V(G) − N[v]]. Then, H has at most one connected component. Otherwise, we can make a vertex in each connected component to change the function values including the vertex v to obtain an RDF of T with weight γdR(T) − 3. Since H contains no vertex assigned with 3, then H is not a star with at least two leaves. We claim that the leaves of H are at most two. Otherwise, we change the function values of v and choose two leaves to obtain an RDF of T with weight γdR(T) − 3. Therefore, H is a path on at least four vertices. In this case, we can obtain an RDF of T with weight at most γdR(T) − 3, a contradiction. Therefore, T is a wounded spider with only one foot.

Conflicts of Interest

The authors declare that there are no conflicts of interest regarding the publication of this paper.

Acknowledgments

This work was supported by Sichuan Science and Technology Program under grant 2018ZR0265, Sichuan Military and Civilian Integration Strategy Research Center under grant JMRH-1818, and Sichuan Provincial Department of Education (Key Project) under grant 18ZA0118.

    Data Availability

    No data were used to support this study.

      The full text of this article hosted at iucr.org is unavailable due to technical difficulties.