332
Views
0
CrossRef citations to date
0
Altmetric
Articles

Lower bounds on nonnegative signed domination parameters in graphs

Abstract

Let 1kn be a positive integer. A nonnegative signed k-subdominating function is a function f:V(G){1,1} satisfying uNG[v]f(u)0 for at least k vertices v of G. The value minvV(G)f(v), taking over all nonnegative signed k-subdominating functions f of G, is called the nonnegative signed k-subdomination number of G and denoted by γksNN(G). If k=|V(G)|, then γksNN(G)=γsNN(G) is the nonnegative signed domination number, introduced in Huang et al. (2013) . In this paper, we obtain several sharp lower bounds of γsNN(G), which extend some known lower bounds on γsNN(G).

We also initiate a study of the nonnegative signed k-subdomination number in graphs and establish some sharp lower bounds for γksNN(G) in terms of the order and the degree sequence of a graph G.

1 Introduction

Let G be a simple graph of order n with vertex set V(G) and size m with edge set E(G). We use Citation[1] for terminology and notation, which are not defined here. If XV(G), then G[X] is the subgraph of G induced by X. For disjoint subsets X and Y of vertices we let E(X,Y) denote the set of edges between X and Y.

The open neighborhood NG(v) of a vertex vV(G) is the set of all vertices adjacent to v. Its closed neighborhood is NG[v]=NG(v){v}. In addition, the open and closed neighborhoods of a subset SV(G) are NG(S)=vSNG(v) and NG[S]=NG(S)S, respectively. The degree of a vertex vV(G) is degG(v)=NG(v). The minimum and maximum degrees in graph G are denoted by δ(G) and Δ(G), respectively. A vertex vV(G) is called an odd (even) vertex if degG(v) is odd (even). For a graph G=(V,E), let Vo ( Ve) be the set of odd (even) vertices with no=Vo and ne=Ve.

For a real-valued function f:V(G)R the weight of f is ω(f)=vV(G)f(v) and for a subset S of V(G) we define f(S)=vSf(v), so ω(f)=f(V(G)). For a positive integer 1kn, a signedk-subdominating function (SkSDF) of G is a function f:V(G){1,1} such that f(NG[v])1 for at least k vertices v of G. The signedk-subdomination number for a graph G is defined as γks(G)=min{w(f)|f is a SkSDF of G}. The signed k-subdomination number was introduced and studied by Cockayne and Mynhardt Citation[2]. In the special case where k=n, γks(G) is the signed domination number γs(G). This concept has been extensively explored in the literature; see e.g. Citation[3–5].

A nonnegative signed dominating function (NNSDF) of G is defined in Citation[6] as a function f:V(G){1,1} such that f(NG[v])0 for all vertices v of G. The nonnegative signed domination number (NNSDN) of G is γsNN(G)=min{ω(f)|f is an NNSDF of G}.

We now introduce a nonnegative signedk-subdominating function (NNSkSDF) of G for a positive integer 1kn as a function f:V(G){1,1} such that f(NG[v])0 for at least k vertices v of G. We define the nonnegative signed k-subdomination number (NNSkSDN) of G by γksNN(G)=min{ω(f)f is a NNSkSDF of G}. A nonnegative signed k-subdominating function of weight γksNN(G) is called a γksNN(G)-function. Note that γnsNN(G)=γsNN(G). Since every signed k-subdominating function of G is a nonnegative signed k-subdominating function, we deduce that γksNN(G)γks(G).For a function f:V(G){1,1} of G we define P={vV(G)f(v)=1}, M={vV(G)f(v)=1}, and Cf={vV(G)|f(NG[v])0}.

In this paper, we establish some new lower bounds on γsNN(G) for a general graph in terms of different graph parameters. Some of these bounds improve several lower bounds on γsNN(G) presented in Citation[6,7]. We also initiate a study of nonnegative signed k-subdomination numbers in graphs, and present some sharp lower bounds for γksNN(G) in terms of the order and the degree sequence of G.

Observation 1

Let f be an NNSDF of G. For vV(G) ifv is an even vertex, then f(NG[v])1 while f(NG[v])0 ifv is an odd vertex.

In this paper, we make use of the following results.

Theorem A

Citation[7] Let G be a graph of order n and size m. Then γsNN(G)n2m.

Theorem B

Citation[7] Let G be a graph of order n, sizem and minimum degree δ. Then γsNN(G)4m+3nδ+12n3δ+12+1.

Theorem C

Citation[4] For n3, γs(Cn)=n3ifn0(mod3),n3+1ifn1(mod3),n3+2ifn2(mod3).

Corollary 2

Citation[8] For any r-regular graph G of order n,γs(G)nr+1, forr even. Furthermore this bound is sharp.

Proposition 3

Citation[6] Let Kn be a complete graph. Then γsNN(Kn)=0 when n is even and γsNN(Kn)=1 when n is odd.

Theorem D

Citation[6] For any graph G with maximum degree Δ and minimum degree δ, we have γsNN(G)δΔδ+Δ+2n.

2 Lower bounds on the NNSDNs of graphs

In this section, we present some new sharp lower bounds for γsNN(G) in terms of ne where ne is the number of even vertices in a graph G. We begin with the following lemma.

Lemma 4

Let f be an NNSDF of a simple connected graph G. Then,

1.

vPdegG(v)n+ne2P+vMdegG(v).

2.

vPdegG[P](v)vPdegG(v)12.

Proof

For vV(G), let degP(v) and degM(v) denote the numbers of vertices of P and M, respectively, which are adjacent to v. Clearly, degG(v)=degM(v)+degP(v). Since f(NG[v])0, for every vP, degM(v)degP(v)+1, and for every vM, degP(v)degM(v)+1. Hence, if vP, then degM(v)degG(v)+12 and if vM, then degP(v)degG(v)+12.

1.

Counting the number of edges in E(P,M) in two ways, we can deduce that vMdegG(v)+12|E(P,M)|vPdegG(v)+12.It follows that vPdegG(v)+PvVdegG(v)+12=vVodegG(v)+12+vVedegG(v)+22=vVdegG(v)2+no2+ne=vPdegG(v)2+vMdegG(v)2+no2+ne, which implies that vPdegG(v)2vMdegG(v)2+no2+neP.Hence, vPdegG(v)vMdegG(v)+n+ne2P.

2.

Consider the subgraph G[P] induced by P. We have degG[P](v)=degP(v) for each vP. Since degP(v)degG(v)12 for each vP, we have vPdegG[P](v)vPdegG(v)12.

In the next theorem we present some lower bounds on γsNN(G). By using Lemma 4 and graph parameters such as order, size, number of even vertices, maximum and minimum degrees we obtain some new lower bounds for γsNN(G).

Theorem 5

Let G be a simple connected graph of order n, minimum degree δ, maximum degree Δ and the number of even vertices ne. Then

1.

γsNN(G)nδnΔ+2neΔ+δ+2,

2.

γsNN(G)2m+nenΔΔ+1,

3.

γsNN(G)nδ+ne2mδ+1,

4.

γsNN(G)(δ+1)+(δ+1)2+8(nδ+n+ne)2n,

5.

γsNN(G)2m+n+nen.

Proof

Let f be a γsNN(G)-function. Let P={vV(G)|f(v)=1} and M={vV(G)|f(v)=1}.

1.

Since δdegG(v)Δ for each vV(G), according to Lemma 4(1) we have δnPδ+n+ne2PvPdegG(v)ΔP.It leads to Pδn+n+neΔ+δ+2.Hence, γsNN(G)=2PnnδnΔ+2neΔ+δ+2.

2.

Since 2m=vVdegG(v) and degG(v)Δ for each vV(G), by Lemma 4(1) it follows that 2m+n+ne2P=vVdegG(v)+n+ne2P2vPdegG(v)2ΔP.Therefore, 2P2m+n+neΔ+1, and γsNN(G)2m+nenΔΔ+1, as desired.

3.

Using Lemma 4(1), the facts 2m=vVdegG(v) and degG(v)δ for any vV(G), we have 2m=vVdegG(v)2vMdegG(v)+n+ne2P2nδ2δP+n+ne2PIt follows that 2P2nδ+n+ne2mδ+1.Thus, γsNN(G)nδ+ne2mδ+1.

4.

Consider G[P] as the induced subgraph of G by P. According to Lemma 4(2), we have vPdegG[P](v)vPdegG(v)12vPdegG(v)12.On the other hand, since G[P] is a simple connected graph, vPdegG[P](v)P(P1).Thus, 2P(P1)vPdegG(v)PvMdegG(v)+n+ne3PnδδP+n+ne3P.This implies that 2P2+(δ+1)P(nδ+n+ne)0.Therefore, 2P(δ+1)+(δ+1)2+8(nδ+n+ne)2, and hence γsNN(G)(δ+1)+(δ+1)2+8(nδ+n+ne)2n, as desired.

5.

By Parts (4) and (2) we have 2vPdegG(v)4P22P,and 2vPdegG(v)2m+n+ne2P,respectively. So, 4P22m+n+ne,It leads to P2m+n+ne2.Thus, γsNN(G)2m+n+nen. This completes the proof. □

From Theorem 5(1) (3), we have the following result.

Corollary 6

For r1, if G is anr-regular graph of order n, then γsNN(G)nr+1 if r is even,0 if r is odd.Furthermore, these bounds are sharp.

For k=n by Observation 1 when r is even, we have the same bound presented in Corollary 2 by Henning. In order to show that the bounds presented in Theorem 5 are sharp, we will give a graph G and construct a γsNN(G)-function f such that w(f) achieves the lower bounds. Our bounds for some of these graphs are attainable while the corresponding bounds given in Theorems A, B, and D are not. In fact, trivial examples such as G{Kn,Cn} are sufficient for this. By Proposition 3 it is easy to see that γsNN(Kn) attains all the five bounds in Theorem 5 while the bound in Theorem A shows that γsNN(Kn)2nn22 and the bound in Theorem B is not more than 5nn23n+5. Moreover, by Theorem C, γsNN(Cn) attains the lower bounds in Theorem 5(1) (3), when n0(mod3) while the bounds in Theorems A, B and D are not more than n7. Besides, we can construct a non-complete graph with an arbitrary large order which attains the lower bounds in Theorem 5(1) (3) as follows. Letting t be a positive integer, we consider a cycle of length 2t. For every edge, we include an additional vertex being adjacent to both endpoints of the corresponding edge. The obtained graph is denoted by G. It is easy to check that G is a graph with n=4t, m=6t, δ=2, Δ=4 and ne=4t. Define a function f:V(G){1,1} as follows: f(v)=1 for vV(C2t) and f(v)=1 for vV(G)V(C2t). It is easy to verify that f is a γsNN(G)-function with w(f)=0 and bounds in Theorem 5(1) (3) are also 0, which implies that G is sharp for these bounds. However, γsNN(G) does not attain the corresponding bounds given in Theorems A, B and D, which are 4t, 4t7, and t, respectively. Next, we show that there is also a graph G different from Kn such that γsNN(G) reaches lower bounds in Theorem 5(4) (5). Let H be the Hajós graph. We can verify that γsNN(H)=0, and H is sharp for presented bounds in Theorem 5(4) (5).

3 Lower bounds on the NNSkSDNs of graphs

In this section, we initiate a study of the nonnegative signed k-subdomination number in graphs. we present some lower bounds on the nonnegative signed k-subdomination number of a graph in terms of the order and the degree sequence. We begin with the following lemma.

Lemma 7

Let G be a graph and 1kn be a positive integer. Let f be a γksNN(G)-function. Let P1=PCf,P2=PP1,M1=MCf andM2=MM1. Then, vPdegG(v)+P1vP1M1degG(v)+12.

Proof

Note that if vV(G), then degG(v)=degP(v)+degM(v). For vP1M1, f(NG[v])0. So, if vP1, then degP(v)degG(v)12 and degM(v)degG(v)+12. Similarly, if vM1, then degP(v)degG(v)+12 and degM(v)degG(v)12.

Counting the number of edges in E(P,M) in two ways, we conclude that vM1degG(v)+12+vM2degP(v)vP1degG(v)+12+vP2degM(v).By adding vP1degG(v)+12 to the both sides of the inequality we have vPdegG(v)+P1vP1M1degG(v)+12+vM2degG(v)vP1M1degG(v)+12,and this completes the proof.□

Theorem 8

For any graph G with degree sequence d1d2dn and any positive integer 1kn,

1.

γksNN(G)2i=1kdi+12Δ+1n.

2.

γksNN(G)nδ4mn+2i=1kdi+12δ+1.

Furthermore, these bounds are sharp.

Proof

Let f be a γksNN(G)-function. Let P1=PCf and M1=MCf.

1.

Since δdegG(v)Δ for each vV(G), Lemma 7 follows that ΔP+PvPdegG(v)+P1vP1M1degG(v)+12.Note that P1M1= and P1M1=P1+M1k. So, PvP1M1degG(v)+12Δ+1i=1kdi+12Δ+1.Thus, γksNN(G)=2Pn2i=1kdi+12Δ+1n.

2.

Obviously, 2m=vV(G)degG(v)=vPdegG(v)+vMdegG(v). If we add P1 to the both sides of this equality, then by Lemma 7 we deduce that PP12m+vP1M1degG(v)+12+vMdegG(v)2m+i=1kdi+12+δnδP.Therefore, Pnδ2m+i=1kdi+12δ+1,and hence, γksNN(G)=2Pnnδ4mn+2i=1kdi+12δ+1.

Now suppose that k=n, considering that 2i=1ndi+12=2m+n+ne, we can immediately obtain those two bounds in Theorem 5(2) and (3) from the lower bounds of Theorem 8, respectively. Since lower bounds presented in Theorem 5 are sharp, so there exist graphs whose γksNN(G) receive lower bounds in Theorem 8. Therefore, these bounds are sharp.□

From Theorem 8(1) or (2) the following result can be immediately deduced.

Corollary 9

For r1, if G is anr-regular graph of order n, then γksNN(G)k(r+2)r+1n if r is even,kn if r is odd.Furthermore, these bounds are sharp.

References

  • WestD.B., Introduction to Graph Theory2000Prentice-Hall, Inc
  • CockayneE.J.MynhardtC.M., On a generalization of signed dominating function of graphs Ars Combin.1996 235–245
  • ChenW.SongE., Lower bounds on several versions of signed domination number Disceret Math. 3082008 1837–1846
  • DunbarJ.HedetniemiS.T.HenningM.A.SlaterP.J., Signed Domination in Graphs, Graph Theory, Combinatorics, and Applications, Vol. 11995WileyNew York311–322
  • HaynesT.W.HedetniemiS.T.SlaterP.J., Domination in Graphs: Advance Topics1998Marcel Dekker, Inc.New York
  • HuangZ.LiW.FengZ.XingH., On nonnegative signed domination in graphs and its algorithmic complexity J. Netw. (2)2013 365–372
  • AtapourM.SheikholeslamiS.M., On the nonnegative signed domination numbers in graphs Electron. J. Graph Theory Appl. (2)2016 231–237
  • HenningM.A., Domination in regular graphs Ars Combin.1996 263–271