468
Views
1
CrossRef citations to date
0
Altmetric
Articles

On [k]-Roman domination subdivision number of graphs

, , &
Pages 261-267 | Received 11 Apr 2022, Accepted 18 Sep 2022, Published online: 24 Oct 2022

Abstract

Let k1 be an integer and G a simple graph with vertex set V(G). Let f be a function that assigns labels from the set {0,1,2,,k+1} to the vertices of G. For a vertex vV(G), the active neighbourhood AN(v) of v is the set of all vertices w adjacent to v such that f(w)1. A [k]-Roman dominating function (or [k]-RDF for short) is a function f:V(G){0,1,2,,k+1} satisfying the condition that for any vertex vV(G) with f(v) < k, f(N[v])|AN(v)|+k. The weight of a [k]-RDF is ω(f)=ΣvV(G)f(v), and the [k]-Roman domination number γ[kR](G) of G is the minimum weight of an [k]-RDF on G. In this paper we shall be interested in the study of the [k]-Roman domination subdivision number sdγ[kR](G) of G defined as the minimum number of edges that must be subdivided, each once, in order to increase the [k]-Roman domination number. We first show that the decision problem associated with sdγ[kR](G) is NP-hard. Then various properties and bounds are established.

MSC 2000:

1. Introduction

For notation and graph theory terminology, we in general follow Haynes, Hedeniemi and Slater [Citation16]. Throughout this paper, G denotes a simple graph, with vertex set V=V(G) and edge set E=E(G). The order |V(G)| of G is denoted by n=n(G). The open neighborhood of a vertex v is the set N(v)={uV(G)|uvE(G)} and the closed neighborhood of a vertex v is the set N[v]=N(v){v}. The degree of a vertex vV is degG(v)=deg(v)=|N(v)|. The minimum and maximum degree of a graph G are denoted by δ(G)=δ and Δ(G)=Δ, respectively. The subdivision of an edge uv of a graph G consists of the deletion of uv from G and the addition of two edges ux and xv along with a new vertex x that will be called a subdivision vertex. As usual, Kn, Pn and Cn denote the complete graph, the path and the cycle of order n.

Let k1 be an integer and let f be a function that assigns labels from the set {0,1,,k+1} to the vertices of G. The active neighborhood AN(v) of a vertex vV(G) with respect to f is the set of all vertices wN(v) such that f(w)1. A [k]-Roman dominating function on a graph G, abbreviated [k]-RDF, is a function f:V(G){0,1,,k+1} satisfying the condition that for any vertex vV(G) with f(v) < k, f(N[v])|AN(v)|+k. The weight of a [k]-RDF is ω(f)=ΣvV(G)f(v), and the [k]-Roman domination number γ[kR](G) of G is the minimum weight of a [k]-RDF on G. A γ[kR](G)-function is a [k]-RDF of weight γ[kR](G). Moreover, if f is a [k]-RDF on G, we let Vif={vV|f(v)=i} for every i{0,1,,k+1}. Consequently, any [k]-RDF f can be represented by f=(V0f,V1f,,Vk+1f), where the superscript f can be deleted in Vif when no confusion arises. The [k]-Roman domination number was introduced by Abdollahzadeh Ahangar at al. in [Citation1] and has been studied by several authors [Citation3, Citation14, Citation15, Citation17, Citation19, Citation20]. Note that γ[1R](G) coincides with the Roman domination number γR(G), γ[2R](G) coincides with the double Roman domination number, γ[3R](G) coincides with the triple Roman domination number, while γ[4R](G) coincides with the quadruple Roman domination number. For more details on Roman domination and its variations we refer the reader to the two book chapters and surveys papers [Citation8–12].

In this paper we are interested in the study of the [k]-Roman domination subdivision number, abbreviated [k]-RDS-number, sdγ[kR](G) of a graph G defined as the minimum number of edges that must be subdivided, where each edge in G is subdivided at most once, in order to increase the [k]-Roman domination number. It is worth mentioning that the [1]-RDS-number matches with the Roman domination subdivision number sdγR(G) which has been studied in [Citation2, Citation4, Citation7, Citation18]. Also, the [2]-RDS-number and [3]-RDS-number coincide with the double Roman domination subdivision and triple Roman domination subdivision numbers, respectively, that have been recently investigated in [Citation5, Citation6].

In this paper, we begin by showing that the decision problem associated with sdγ[kR](G) is NP-hard, and then upper bounds on the [k]-RDS-number are established for arbitrary graphs.

Before further proceeding, it should be noted that since the [k]-RDS-number of the graph K2 does not change when its only edge is subdivided, we will assume that at least one component of the graph G has order at least 3. Moreover, if G1,G2,,Gs are the components of G then γ[kR](G)=Σi=1sγ[kR](Gi) and if G1,G2,,Gt are the components of G of order at least 3, then sdγ[kR](G)=min{sdγ[kR](Gi)|1it}. Hence, we will only consider connected graphs of order at least three.

We start by showing that the subdivision of any edge does not decrease the [k]-Roman domination number.

Proposition 1.

Let G be a simple connected graph of order n3 and e=uvE(G). If G is obtained from G by subdivision the edge e, then γ[kR](G)γ[kR](G) for every integer k1.

Proof.

Let x be the subdivision vertex and let f be a γ[kR](G)-function. Define the function g:V(G){0,1,,k+1} by g(u)=min{k+1,f(u)+f(x)} and g(z)=f(z) for zV(G){u}. Clearly g is a [k]-RDF on G and ω(g)ω(f). Hence γ[kR](G)γ[kR](G) as desired.□

We end this section by the following results that are useful to our proofs.

Proposition 2

([Citation17]). If k2, then in a [k]-RDF of weight γ[kR](G), no vertex needs to be assigned the value 1.

Observation 3.

Let G be a connected graph of order n3. Then γ[kR](G)k+1, with equality if and only if Δ(G)=n1.

2. NP-hardness result

In this section, we will show that the [k]-Roman domination subdivision number problem, to which we shall refer as [k]-RDSN, is NP-hard even for bipartite graphs.

[k]-Roman domination subdivision number problem ([k]-RDSN):

Instance: A nonempty bipartite graph G and a positive integer l.

Question: Is sdγ[kR](G)l?

Following Garey and Johnson’s techniques for proving NP-completeness given in [Citation13], we prove our result by describing a polynomial transformation from the well-known NP-complete problem 3-SAT to [k]-RDSN. Before stating the 3-SAT problem, let us recall the following.

Let U be a set of Boolean variables. A truth assignment for U is a mapping t:U{T,F}. If t(u) = T, then u is said to be “true” under t; if t(u) = F, then u is said to be “false” under t. If u is a variable in U, then u and u¯ are literals over U. The literal u is true under t if, and only if, the variable u¯ is false under t; the literal u¯ is true if, and only if, the variable u is false.

A clause over U is a set of literals over U. It represents the disjunction of these literals and is satisfied by a truth assignment if, and only if, at least one of its members is true under that assignment. A collection C of clauses over U is satisfiable if, and only if, there exists some truth assignment for U that simultaneously satisfies all the clauses in C. Such a truth assignment is called a satisfying truth assignment for C. The 3-SAT is specified as follows.

3. Satisfiability problem (3-SAT)

Instance: A collection C={C1,C2,,Cm} of clauses over a finite set U of variables such that |Cj|=3 for j=1,2,,m.

Question: Is there a truth assignment for U that satisfies all the clauses in C?

Theorem 4

([Citation13] Theorem 3.1). 3-SAT is NP-complete.

Theorem 5.

The [k]-RDSN problem is NP-hard for bipartite graphs for every integer k2.

Proof.

Let U={u1,u2,,un} and C={C1,C2,,Cm} be an arbitrary instance of 3SAT. We will construct a bipartite graph G and choose an integer l such that C be satisfiable if and only if sdγ[kR](G)l. We construct such a bipartite graph G as follows.

For each i{1,2,,n}, we associate to each variable uiU, a copy of the complete bipartite graph Hi=K3,5 with bipartite sets X={xi,yi,zi} and Y={vi,ui,wi,ui¯,ri}. For each j{1,2,,m}, we associate to each clause CjC, a single vertex cj by adding the edges cjui if uiCj and cjui¯ if ui¯Cj. Finally, to obtain the graph G, we add a graph H isomorphic to a path P10:s1s2s9s10 by joining each of s1 and s10 to all cj’s. Set l=1.

provides an example of the constructed graph when U={u1,u2,u3,u4} and C={C1,C2,C3}, where C1={u1,u2,u3¯},C2={u1¯,u2,u4},C3={u2¯,u3,u4}. To prove that this is indeed a transformation, we only need to show that sdγ[kR](G)=1 if and only if there is a truth assignment for U that satisfies all clauses in C. This aim will be achieved through the proof of the following four claims.

Figure 1. An instance of the [k]-Roman subdivision number problem resulting from an instance of 3SAT. Here l=1 and γ[kR](G)=12k+11, where the bold vertex p means there is a [k]-RDF f with f(p)=k+1, with the exception of s4 where f(s4)=k.

Figure 1. An instance of the [k]-Roman subdivision number problem resulting from an instance of 3SAT. Here l=1 and γ[kR](G)=12k+11, where the bold vertex p means there is a [k]-RDF f with f(p)=k+1, with the exception of s4 where f(s4)=k.

Claim 1. γ[kR](G)(2n+4)(k+1)1. Moreover, if γ[kR](G)=(2n+4)(k+1)1, then for any γ[kR]-function f on G, f(V(Hi))=2(k+1),f(s1)=f(s10)=0 or {f(s1),f(s2)}={0,k} and f(cj)=0 for each 1jm.

Proof

of Claim 1. Let f be a γ[kR]-function of G. Without loss of generality, assume that f(z)1 for all vertex zV(G) (by Proposition 2). For each i{1,2,,n}, it is clear that f(V(Hi))2(k+1) and f(V(H))+i=1mf(cj)4k+3, implying that γ[kR](G)(2n+4)(k+1)1. Now suppose that γ[kR](G)=(2n+4)(k+1)1. Since each Hi is isomorphic to K3,5,f(V(Hi))=2(k+1). Next, we show that f(s1)=f(s10)=0 or {f(s1),f(s10)}={0,k}. Suppose for a contradiction that f(s1)0 or f(s10)0 and {f(s1),f(s10}{0,k}.

If f(s1)=k+1 (the case f(s10)=k+1 is similar), then since f(NG[s4])k+1,f(NG[s7])k+1 and thus f(NG[s10])k+1 implying that γ[kR](G)=(2n+4)(k+1), which leads to a contradiction. Hence let 0<f(s1)<k (the case 0<f(s10)<k is similar), then f(NG[s1])k+1 and f(NG[s4])+f(NG[s7])+f(s9)+f(s10)3(k+1), implying that γ[kR](G)=(2n+4)(k+1), a contradiction. Therefore f(s1)=f(s10)=0 or {f(s1),f(s10}={0,k}. It follows that f(V(H))=4k+3 and thus j=1mf(cj)=0, as desired.

Claim 2. C is satisfiable if and only if γkR(G)=(2n+4)(k+1)1.

Proof

of Claim 2. Suppose that γ[kR](G)=(2n+4)(k+1)1 and let f be a γ[kR]-function of G that assigns no 1 to any vertex of G. By Claim 1, f(V(Hi))=2(k+1) for each i{1,2,.n}, and since Hi is a complete bipartite graph, at most one of f(ui) and f(ui¯) is assigned k + 1 for each i. Define the mapping t:U{T,F} by (1) t(ui)={Tif either f(ui)=k+1 or f(ui)k+1 and f(u¯i)k+1Fif f(ui¯)=k+1,(1) for each i{1,,n}. We now show that t is a satisfying truth assignment for C. It is sufficient to show that every clause in C is satisfied by t. To this end, we arbitrarily choose a clause CjC for some j{1,,m}.

By Claim 1, since f(s1)+f(s10)<k+1 and cjshE(G) for every h{2,,9}, there exists some index i with i{1,,n} such that cj is adjacent to ui or ui¯. Suppose that cj is adjacent to ui where f(ui)=k+1. Since ui is adjacent to cj in G, the literal ui is in the clause Cj by the construction of G. Since f(ui)=k+1, it follows that t(ui)=T by (1), which implies that the clause Cj is satisfied by t. Suppose that cj is adjacent to ui¯ where f(ui¯)=k+1. Since ui¯ is adjacent to cj in G, the literal ui¯ is in the clause Cj. Since f(ui¯)=k+1, it follows that t(ui)=F by (1). Thus, t assigns ui¯ the truth value T, that is, t satisfies the clause Cj. By the arbitrariness of j, we conclude that t satisfies all the clauses in C, so C is satisfiable.

Conversely, suppose that C is satisfiable, and let t:U{T,F} be a satisfying truth assignment for C. Create a function f on V(G) as follows: if t(ui)=T, then let f(ui)=k+1, and if t(ui)=F, then let f(ui¯)=k+1. Let f(yi)=k+1 for each i, f(s2)=f(s6)=f(s9)=k+1,f(s4)=k, and the remaining vertices of G are assigned a 0 under f. Clearly, f(G)=(2n+4)(k+1)1. Since t is a satisfying truth assignment for C, for each j{1,2,,m}, at least one of literals in Cj is true under the assignment t. It follows that the corresponding vertex cj in G is adjacent to at least one vertex p with f(p)=k+1. Since cj is adjacent to each literal in Cj by the construction of G, thus f is a [k]-RDF of G, and so γ[kR](G)f(G)=(2n+4)(k+1)1. By Claim 1, γ[kR](G)(2n+4)(k+1)1, and thus γ[kR](G)=(2n+4)(k+1)1.

Claim 3. Let G be obtained from G by subdividing any edge e of E(G), then γ[kR](G)(2n+4)(k+1).

Proof

of Claim 3. Let e=uvE(G) and let G be obtained from G by subdividing the edge e. Consider the following cases:

Case 1.

e=sisi+1 for some i{1,,9}.

Without loss of generality, let e=s1s2 and define the function f on V(G) by f(s1)=f(s3)=f(s6)=f(s9)=f(xi)=f(vi)=k+1 for each 1in and f(x) = 0 for all other xV(G).

Case 2.

e{s1cj,s10cj}, for some j{1,,m}.

Consider the function f defined on V(G) by f(s1)=f(s4)=f(s7)=f(s10)=f(xi)=f(vi)=k+1 for each 1in and f(x) = 0 for all other xV(G).

Case 3.

e=cjui, for some i{1,,n} and some j{1,,m} or e=cju¯i, for some i{1,,n} and some j{1,,m}.

Consider the function f defined on V(G) by f(s1)=f(s4)=f(s7)=f(s9)=f(xi)=f(ui)=k+1 or f(s1)=f(s4)=f(s7)=f(s9)=f(xi)=f(ui¯)=k+1 respectively, for each i{1,,n} and f(x) = 0 for all other xV(G).

Case 4.

e = uv where u{xi,yi,zi} and v{vi,ui,wi,u¯i,ri}.

Consider the function f defined on V(G) by f(s1)=f(s4)=f(s7)=f(s9)=f(u)=f(v)=k+1 and f(x) = 0 for all other xV(G).

In either case, f is a [k]-RDF on G of weight (2n+4)(k+1), and therefore γ[kR](G)(2n+4)(k+1).

Claim 4. γ[kR](G)=(2n+4)(k+1)1 if and only if sdγ[kR](G)=1.

Proof

of Claim 4. Assume γ[kR](G)=(2n+4)(k+1)1, and let G be the graph obtained from G by subdivision the edge e=s1s2 with vertex subdivision w.

Suppose for a contradiction that γ[kR](G)=γ[kR](G), and let f be a γ[kR](G)-function. Define the function g on V(G) by g(s1)=min{f(s1)+f(w),k+1} and g(z)=f(z) for zV(G){s1}. Clearly, g is a [k]-RDF of G and thus γ[kR](G)ω(g)ω(f)=γ[kR](G), implying that γ[kR](G)=ω(g) and g is a γ[kR](G)-function on G. By Claim 1, g(s1)=0 or g(s1)=k. If g(s1)=0, then f(s1)+f(w)=0 and so f(s1)=f(w)=0. It follows that j=1mf(cj)0, and since g(z)=f(z) for zV(G){s1}, we deduce that j=1mg(cj)0, which leads to a contradiction (with Claim 1). If g(s1)=k, then f(s1)+f(w)=k, and since f(s1)+f(w)+j=1mf(cj)k+1, we deduce that j=1mf(cj)0, yielding as above to the contradiction j=1mg(cj)0. Therefore, γ[kR](G)<γ[kR](G), and thus sdγ[kR](G)=1.

Now, assume that sdγ[kR](G)=1, and let G be the graph obtained from G by subdividing the edge e for which γ[kR](G)<γ[kR](G). By Claim 3, we have γ[kR](G)(2n+4)(k+1). It follows from Claim 1 that (2n+4)(k+1)1γ[kR](G)<γ[kR](G)(2n+4)(k+1), and thus γ[kR](G)=(2n+4)(k+1)1, as desired.

By Claims 2 and 4, we showed that sdγ[kR](G)=1 if and only if there is a truth assignment for U that satisfies all clauses in C. Since the construction of the [k]-RDSN instance is straightforward from a 3-satisfiability instance, the size of the [k]-RDSN instance is bounded above by a polynomial function of the size of 3-satisfiability instance. It follows that this is a polynomial reduction and the proof is complete.□

4. Bounds on the [k]-RDS-number

In this section we present some sufficient conditions for graphs to have small [k]-RDS-number. Recall that a vertex of degree one is called a leaf and its neighbor a support vertex.

Proposition 6.

If G contains a support vertex with at least two leaf neighbors, then for every integer k1,sdγ[kR](G)=1.

Proof.

Let u, v be two leaves sharing the same neighbor w and let G be the graph obtained from G by subdividing the edge uw with vertex subdivision x. If f is any [k]-RDF on G, then we must have f(u)+f(x)k and f(w)+f(v)k. In this case, let us define the function g on V(G) by g(w)=k+1,g(u)=g(v)=0 and g(z)=f(z) for each zV(G){u,v,w}. Clearly, g is a [k]-RDF of G such that ω(g)<ω(f) and hence sdγ[kR](G)=1.

Proposition 7.

Let G be a connected graph of order n3. Then for every integer k1, if γ[kR](G)=k+1, then sdγ[kR](G)=1.

Proof.

Assume that γ[kR](G)=k+1. By Observation 3, Δ(G)=n1. Let v be a vertex with maximum degree Δ(G) and let wN(v). Let G be the graph resulting from G by subdividing the edge vw with vertex x. Then Δ(G)<n(G)1=n and so γ[kR](G)k+2 by Observation 3, which leads to sdγ[kR](G)=1.

Proposition 8.

Let G be a connected graph of order n3 with δ(G)=1. Then for every integer k2,sdγ[kR](G)2.

Proof.

Let uV(G) be a leaf, v its support vertex and w a neighbor of v different from u. Clearly, if w is a leaf, then by Proposition 6, the result is valid. Hence we assume that w is not a leaf. Let G be the graph resulting from G by subdividing the edges uv, vw with subdivision vertices x and y, respectively. Let f be a γ[kR](G)-function such that f(u) is as small as possible. By Proposition 2 we may assume that f(z)1 for all vertex zV(G). First suppose that f(x) = 0. By the choice of f we must have f(u) = k and we deduce from f[N(x)]k+|AN(x)| that f(v)2. But then the function g defined by g(w)=min{f(w)+f(y),k+1}, g(v)=k+1, g(u) = 0 and g(z)=f(z) otherwise, is a [k]-RDF of G of weight less that ω(f) yielding the desired result.

Assume now that 0<f(x)k. It follows from f[N(u)]k+|AN(u)| that f(u)+f(x)k+1. But then the function h defined by h(u) = 0, h(x)=k+1 and h(z)=f(z) otherwise, is a [k]-RDF of G of weight ω(f) which contradicts the choice of f.

Finally, let f(x)=k+1. Then we have f(u) = 0. If f(v)1, then the function g defined as before, is a [k]-RDF of G of weight less that ω(f) as desired. Hence assume that f(v) = 0. If f(y) = 0, then f(w)=k+1 and the function h defined on G by h(u) = k and g(z)=f(z) otherwise, is a [k]-RDF of G of weight less that ω(f). If f(y)1, then the function l defined on G by l(v)=k+1 and l(z)=f(z) otherwise, is a [k]-RDF of G of weight less than ω(f). In either case, sdγ[kR](G)2.

Proposition 8 leads to the following corollary.

Corollary 9.

Let T be a tree of order at least 3. Then for every integer k2, sdγ[kR](T)2.

Furthermore, this bound is sharp for paths of order n with n2(mod3).

Theorem 10.

Let G be a connected graph and let v be a vertex of degree at least two. Then for every integer k2,sdγ[kR](G)deg(v).

Proof.

Let t=deg(v),N(v)={v1,v2,,vt} and let G be the graph obtained from G by subdividing the edges vv1,vv2,,vvt with new vertices x1,x2,,xt, respectively. Let f be a γ[kR](G)-function. If f(v)+i=1tf(xi)k+2, then the function g defined on V(G) by g(v)=k+1 and g(z)=f(z) for zV(G){v}, is a [k]-RDF of G with ω(g)<ω(f) and hence sdγ[kR](G)deg(v). Hence we can assume that f(v)+i=1tf(xi)k+1. Recall that f is a [k]-RDF on G and so f(N[v])k+|AN(v)|. If f(v)=k+1, then let g be a function defined on V(G) by g(v) = k and g(z)=f(z) otherwise. Clearly, g is a [k]-RDF of G with ω(g)<ω(f) and hence sdγ[kR](G)deg(v). Assume now that f(v) = k. Then by Proposition 2 and our earlier assumption we must have i=1tf(xi)=0 and thus f(vi)2 for each i{1,,t}. Since t2, the function g defined on V(G) by g(v)=k1 and g(x)=f(x) for xV(G){v} is a [k]-RDF of G with ω(g)<ω(f). Finally, assume that 2f(v)k1. Then we deduce from i=1tf(xi)+f(v)k+1 and f(N[v])k+|AN(v)| that there exists some xi, say without loss of generality, x1, such that f(v)+f(x1)=k+1. In this case, define the function g on V(G) by g(v) = k and g(z)=f(z) for each zV(G){v}. Clearly, g is a [k]-RDF of G with ω(g)<ω(f) and hence sdγ[kR](G)deg(v).

Now assume that f(v) = 0. Then as above we must have f(xi)=k+1 for some i{1,,t}, say i = 1 and f(xj)=0 for all j{2,,t}. Hence f(vi)=k+1 for every i{2,,t}. In that case, since t2, the function g defined on V(G) by g(v1)=k and g(z)=f(z) for zV(G){v1} is a [k]-RDF of G with ω(g)<ω(f). Therefore γ[kR](G)ω(g)<ω(f) implying that sdγ[kR](G)deg(v).

Based on Theorem 10, we deduce that sdγ[kR](G) is defined for every connected graph G of order n3. In addition, we have the following consequences, one of which follows from the fact that every planar graph contains a vertex of degree at most 5.

Corollary 11.

For every connected graph G with δ2 and k2 an integer, sdγ[kR](G)δ.

Corollary 12.

For every planar graph G and k2 an integer, sdγ[kR](G)5.

In the sequel, we present an upper bound on the [k]-RDS-number in terms of δ2 defined as follows. For a vertex vV, let N2(v) denote the set of vertices of G that are at distance 2 from v, and let d2(v)=|N2(v)|. In that case, δ2(G)=min{d2(v)|vVanddegG(v)2}. We make use of the following lemmas in the proof of Theorem 17. Recall that the complete graph K3 is called a triangle.

Lemma 13.

Let G be a connected graph of order n3 and let G have a vertex vV(G) which is contained in a triangle uvw such that N(u)N(w)N[v]. Then for every integer k2,sdγ[kR](G)3.

Proof.

Let G be the graph obtained from G by subdividing the edges vu, vw, uw with new vertices x, y, z respectively, and let g be a γ[kR](G)-function. By Proposition 2 we may assume that g(t)1 for all vertices tV(G). We claim that g(u)+g(v)+g(w)+g(x)+g(y)+g(z)k+2. First assume that 0{f(x),f(y),f(z)}, say, without loss of generality, f(x) = 0. Then we must have f(N[x])k+|AN(x)|. If 0{f(u),f(v)}, then clearly f(u)+f(v)k+2 as desired. Hence we can assume that 0{f(u),f(v)}. If f(u) = 0 (the case f(v) = 0 is similar), then f(v)=k+1, and thus we must have for vertex z, f(z)+f(w)k which leads to the desired claim.

Now assume that 0{f(x),f(y),f(z)}. If one of x, y, z is assigned k + 1, then the claim is valid. If f(x)=f(y)=f(z)=k, then again, we are done. Hence assume, without loss of generality, that 1<f(x)<k. By definition, we have f(x)+f(u)+f(v)k+1 and since f(y)1 the claim follows.

Define the function h on V(G) by h(v)=k+1,h(u)=h(w)=0 and h(s)=f(s) for each sV(G){v,u,w}. Obviously h is a [k]-RDF of G of weight less than γ[kR](G), and thiscompletes the proof.□

Lemma 14.

Let G be a connected graph of order n3 and let G have a vertex vV(G) which is contained in a triangle uvw such that N(u)N[v] and N(w)N[v]. Then for every integer k2, sdγ[kR](G)3+|N(w)N[v]|.

Proof.

Let N(w)N[v]={w1,w2,,wt} and let G be the graph obtained from G by subdividing the edges vu, vw, uw with new vertices x, y, z respectively, and by subdividing each edge wwi with a new vertex zi. Let g be a γ[kR](G)-function. As seen in the proof of Lemma 13, we have g(v)+g(u)+g(w)+g(x)+g(y)+g(z)k+2. Define h:V(G){0,1,2,,k+1} by h(v)=k+1,h(u)=h(w)=0,h(wi)=min{g(wi)+g(zi),k+1} for each i{1,,t} and h(s)=g(s) for any remaining vertex s. It is easy to see that h is a [k]-RDF of G of weight less than γ[kR](G) and the proof is complete.□

Lemma 15.

Let G be a connected graph of order n3 and v a vertex of G of degree at least 2 such that

  1. N(y)N[v] for each yN(v),

  2. there exists a pair of vertices α, β in N(v) such that (N(α)N(β))N[v]=.

Then for every integer k2,sdγ[kR](G)3+|N2(v)|.

Proof.

Let N(v)={v1,v2,,vdeg(v)}. Without loss of generality, assume that α=v1 and β=v2. Moreover, we will assume that the pair α, β is chosen first among the adjacent vertices in N(v). Hence if αβE(G), then we assume also that v1v2E(G). In addition, let S={v1,v2,,vt} be one of the largest subset of N(v) containing v1, v2 and such that every pair x, y of vertices of S satisfies (ii). According to item (i), let N(vi)N[v]={vi1,vi2,,vili} for each i{1,2,,t}. Now consider the graph G1 obtained from G by subdividing the edges vv1 and vv2 with new vertices x1 and x2, respectively, and for all i{1,2,,t}, each of the edges vivij,1jli, with a new vertex vij. We put Ti={vij|1jli} and T=1itTi. Furthermore, if v1 and v2 are adjacent, then we also subdivide the edge v1v2 with a vertex u. Let f be a γ[kR](G1)-function such that V1=. Assume first that v1v2E(G). As in the proof of Lemma 13 we can see that f(u)+f(v)+f(v1)+f(v2)+f(x1)+f(x2)k+2. Then the function g defined on V(G) by g(v)=k+1,g(v1)=g(v2)=0,g(vij)=min{f(vij)+f(vij),k+1} for all i{1,2,,t} and all j{1,2,,li} and g(x)=f(x) otherwise, is [k] RDF of G of weight less than γ[kR](G1).

Assume now that v1v2E(G). By the choice of v1, v2, we conclude that S is an independent set. Also to dominate x1, x2, we must have f(v)+f(x1)+f(x2)+f(v1)+f(v2)k+1. If f(v)+f(x1)+f(x2)+i=1tf(vi)k+2, then the function g defined on V(G) by g(v)=k+1,g(vi)=0 for i{1,2,,t},g(vij)=min{f(vij)+f(vij),k+1} for all i{1,2,,t} and all j{1,2,,li} and g(x)=f(x) otherwise, is a [k]-RDF of G of weight less than γ[kR](G1). Thus we may assume that f(v)+f(x1)+f(x2)+i=1tf(vi)=k+1. It follows that f(v)=k+1, otherwise we must have f(x1)+f(x2)+f(v1)+f(v2)k+2, which is a contradiction. Hence f(x1)+f(x2)+i=1tf(vi)=0. To [k]-Roman dominate vij we must have f(vij)+f(vij)k for all i{1,2,,t} and all j{1,2,,li}. Define the function g on V(G) by g(v) = k, g(vij)=min{f(vij)+f(vij),k+1} for all i{1,3,,t} and all j{1,2,,li}, and g(x)=f(x) otherwise. Since each vertex of N(v)S has a common neighbor with vi in N2(v) for some i{1,,t}, the function g is a [k]-RDF of G of weight less than γ[kR](G1). In each of the previous situations we saw that the graph G has a [k]-RDF of weight less than γ[kR](G1). Moreover, since G1 was obtained by inserting at most 3+|T|3+|N2(v)| new vertices, we deduce that sdγ[R](G)3+|N2(v)|. This complete the proof.□

Lemma 16.

Let G be a connected graph of order n3 and let v be a vertex of G of degree at least 2 such that

  1. N(y)N[v] for each yN(v)

  2. for every pair of vertices α, β in N(v),(N(α)N(β))N[v].

Then for every integer k2, sdγ[kR](G)3+|N2(v)|.

Proof.

If deg(v)3+|N2(v)|, then the result is immediate from Theorem 10. Hence assume that deg(v)4+|N2(v)|, and let N(v)={v1,v2,,vt} and set X1=N(v1)N[v]={w1,w2,,wp}. By item (ii), each yN(v){v1} has a neighbor in X1. Let T be one of the largest subset of N(v){v1} such that for each T1T we have |N(T1)(N[v]X1)||T1|. Note that such a set T may not exist, but the inequality |N2(v)||X1|+|T| is still verified. Moreover, every vertex u in U=N(v)(T{v1}) has a neighbor in X1 and N(u)N[v]X1N(T). In addition X1 dominate N(v) (by item (ii)). It follows from 4+|X1|+|T|4+|N2(v)|deg(v)=|T|+1+|U| that |U|4. If T, then, without loss of generality, let T={v2,v3,,vs}.

Let G1 be the graph obtained from G by subdividing the edges v1wj with new vertices yj for all j{1,2,,p} and the edges vvi with new vertices xi for i{1,,s+2} when T or i{1,2,3} when T=. Therefore |X1|+|T|+3 edges of G are subdivided. Recall that |X1|+|T|+3|N2(v)|+3. Let f be a γ[kR](G1)-function such that V1=. If f(v)+f(v1)+i=1s+2f(xi)k+2, then the function g defined on V(G) by g(v)=k+1,g(wj)=min{k+1,f(wj)+f(yj)|1jp} and g(x)=f(x) otherwise, is [k]-RDF of G of weight less than γ[kR](G), yielding the result. Hence we assume that (2) f(v)+f(v1)+i=1s+2f(xi)k+1.(2)

If f(v) = k, then to [k]-Roman dominate x1 we must have f(x1)+f(v1)2 which contradicts (2). Hence f(v)k. Consider the following cases depending on the values of f(v).

Case 1.

f(v)=k+1.

It follows from (2) that f(v1)=i=1s+2f(xi)=0. If i=1pf(yi)k+2, then the function g defined on V(G) by g(v1)=k+1, and g(x)=f(x) otherwise, is a [k]-RDF of weight less than γ[kR](G1). Thus we may assume that i=1pf(yi)k+1. Observe that if f(yj)=0 for some j, then since f(v1)=0, we deduce that f(wj)=k+1. Moreover, if 2f(yj)k1, then f(wj)+f(yj)k+1. Using the fact that each vi has a neighbor in X1, it follows that the function g defined on V(G) by g(v)=k,g(wj)=min{k+1,g(yj)+g(wj)} for each j{1,2,,p} and g(x)=f(x) otherwise is an [k]-RDF of G of weight less than γ[kR](G1). In either case, the desired result follows.

Case 2.

f(v) = 0.

To [k]-Roman dominate x1, we must have f(x1)+f(v1)k. On the other hand, f(x1)+f(v1)k+1 by (2). We deduce from V1= that f(xi)=0 for each 2is+2 and so f(vi)=k+1 for each i{2,,s+2}. In this case, define the function g on V(G) by g(v1)=g(v)=k+1,g(vs+1)=g(vs+2)=0 and g(x)=f(x) otherwise. Since N(vi)N[v]X1N(T) for i{s+1,s+2}, one can easily seen that g is a [k]-RDF of G of weight less than γ[kR](G1) leading to the desired result.

Case 3.

f(v){2,,k1}.

Let f(v)=l. If f(v1)1, then we must have f(v1)+f(x1)kl+2 which contradicts (2). Hence assume that f(v1)=0. Then we must have f(x1)kl+1, and thus the function g defined on G by g(v) = k, g(wi)=min{f(wi)+f(yi),k+1} for 1ip, and g(x)=f(x) otherwise, is a [k]-RDF of G of weight less than γ[kR](G1). In either case, γ[kR](G)<γ[kR](G1), and this completes the proof.□

We are now ready to state and prove the main result of this section.

Theorem 17.

Let G be a connected graph of order n3. Then for every integer k2, sdγ[kR](G)3+min{d2(v):vVanddeg(v)2}.

Proof.

If G is a star K1,n1, then sdγ[kR](G)=1, and thus the result is valid. Hence assume that G is not a star. If G has a leaf, then by Proposition 8, the result is valid again. Hence we assume that G is a graph with δ(G)2. According to Lemmas 13, 14, 15 and 16 we have sdγ[kR](G)3+min{d2(v);vVanddeg(v)2}.

The following two corollaries are immediate consequences of Theorem 17. Notice that for a vertex v of degree Δ, |N2(v)|nΔ1.

Corollary 18.

Let G be a connected graph of order n3. Then for every integer k2, sdγ[4R](G)δ2(G)+3.

Corollary 19.

Let G be a connected graph of order n3. Then for every integer k2,sdγ[kR](G)nΔ+2.

Theorem 20.

Let G be a connected graph of order n3. Then for every integer k2,sdγ[kR](G)n2+1.

Proof.

If G is a star K1,n1, then sdγ[kR](G)=1, and thus the result is valid. Hence assume that G is not a star. If G has a leaf, then by Proposition 8, the result is valid again. Hence we assume that G is a graph with δ(G)2. If δ(G)n2+1, then the result follows from Corollary 11. Assume that δ(G)n2+2. Corollary 19 leads to sdγ[kR](G)n2 as desired.□

References

  • Abdollahzadeh Ahangar, H., Álvarez, M. P., Chellali, M., and Sheikholeslami, S. M, Valenzuela-Tripodoro, J. C. (2021). Triple Roman domination in graphs. Appl. Math. Comput. 391: 125444.
  • Amjadi, J. (2020). Total Roman domination subdivision number in graphs. Commun. Comb. Optim. 5: 157–168.
  • Amjadi, J., Khalili, N. (2022). Quadruple Roman domination in graphs. Discrete Math. Algorithms Appl. 14: 2150130.
  • Amjadi, J., Khoeilar, R., Chellali, M, Shao, Z. (2020). On the Roman domination subdivision number of a graph. J. Comb. Optim. 40(2): 501–511.
  • Amjadi, J., Sadeghi, H. (2022). Double Roman domination subdivision number in graphs. Asian-Eur. J. Math. 15: 2250125.
  • Amjadi, J., Sadeghi, H. (2022). Triple Roman domination subdivision number in graphs. Comput. Sci. J. Moldova 30(1): 109–130.
  • Atapour, M., Sheikholeslami, S. M., Khodkar, A. (2009). Roman domination subdivision number of a graphs. Aequat. Math. 78(3): 237–245.
  • Chellali, M., Jafari Rad, N., Sheikholeslami, S. M., Volkmann, L. (2020). Roman domination in graphs. In Topics in Domination in Graphs, Haynes, T. W.; Hedetniemi, S. T.; Henning, M. A., Eds.; Springer: Berlin/Heidelberg, Germany, pp. 365–409.
  • Chellali, M., Jafari Rad, N., Sheikholeslami, S. M., Volkmann, L. (2020). Varieties of Roman domination. In Structures of Domination in Graphs, Haynes, T. W.; Hedetniemi, S. T.; Henning, M. A., Eds.; Springer Berlin/Heidelberg, Germany, pp. 273–307.
  • Chellali, M., Jafari Rad, N., Sheikholeslami, S. M., Volkmann, L. (2020). Varieties of Roman domination II. AKCE Int. J. Graphs Comb. 17(3): 966–984.
  • Chellali, M., Jafari Rad, N., Sheikholeslami, S. M., Volkmann, L. (2020). The Roman domatic problem in graphs and digraphs: A survey. Discuss. Math. Graph Theory 42(3): 861–891.
  • Chellali, M., Jafari Rad, N., Sheikholeslami, S. M., Volkmann, L. (2020). A survey on Roman domination parameters in directed graphs. J. Combin. Math. Combin. Comput. 115: 141–171.
  • Garey, M. R., Johnson, D. S. (1979). Computers and Intractability: A Guide to the Theory of NP-Completness; San Francisco, CA: Freeman.
  • Hajjari, M., Abdollahzadeh Ahangar, H., Khoeilar, R., Shao, Z., Sheikholeslami, S. M. (2022). New bounds on the triple Roman domination number of graphs. J. Math. 2022: 1–5.
  • Hajjari, M., Abdollahzadeh Ahangar, H., Khoeilar, R., Shao, Z., Sheikholeslami, S. M. An upper bound on triple Roman domination. Commun. Comb. Optim. (In press)
  • Haynes, T. W., Hedetniemi, S. T., Slater, P. J. (1998). Fundamentals of Domination in Graphs; New York, NY: Marcel Dekker, Inc.
  • Khalili, N., Amjadi, J., Chellali, M., Sheikholeslami, S. M. [k]-Roman domination in graphs. Submitted.
  • Khodkar, A., Mobaraky, B. P., Sheikholeslami, S. M. (2013). Roman dominatiom subdivision number of a graph and its complement. Ars. Combin. 111: 97–10.
  • Nahani Pour, F., Abdollahzadeh Ahangar, H., Chellali, M., Sheikholeslami, S. M. (2022). Global triple Roman dominating function. Discrete Appl. Math. 314: 228–237.
  • Poureidi, A., Fathali, J. Algorithmic complexity of triple Roman dominating functions on graphs. Commun. Comb. Optim. (to appear).