
A signed double Italian dominating function (SDIDF) on a graph G=(V,E) is a function f from V to {−1,1,2,3}, satisfying (ⅰ) ∑u∈N[v]f(u)≥1 for all v∈V; (ⅱ) if f(v)=−1 for some v∈V, then there exists A⊆N(v) such that ∑u∈Af(u)≥3; and (ⅲ) if f(v)=1 for some v∈V, then there exists A⊆N(v) such that ∑u∈Af(u)≥2. The weight of an SDIDF f is ∑v∈Vf(v). The signed double Italian domination number of G is the minimum weight of an SDIDF on G. In this paper, we initiated the study of signed double Italian domination and proved that the decision problem associated with the signed double Italian domination is NP-complete. We also provided tight lower and upper bounds of a signed double Italian domination number of trees, and we characterized the trees achieving these bounds. Finally, we determined the signed double Italian domination number for some well-known graphs.
Citation: Ahlam Almulhim. Signed double Italian domination[J]. AIMS Mathematics, 2023, 8(12): 30895-30909. doi: 10.3934/math.20231580
[1] | Fu-Tao Hu, Xing Wei Wang, Ning Li . Characterization of trees with Roman bondage number 1. AIMS Mathematics, 2020, 5(6): 6183-6188. doi: 10.3934/math.2020397 |
[2] | Saeed Kosari, Yongsheng Rao, Zehui Shao, Jafar Amjadi, Rana Khoeilar . Complexity of signed total k-Roman domination problem in graphs. AIMS Mathematics, 2021, 6(1): 952-961. doi: 10.3934/math.2021057 |
[3] | Abel Cabrera Martínez, Iztok Peterin, Ismael G. Yero . Roman domination in direct product graphs and rooted product graphs. AIMS Mathematics, 2021, 6(10): 11084-11096. doi: 10.3934/math.2021643 |
[4] | Zhibin Du, Ayu Ameliatul Shahilah Ahmad Jamri, Roslan Hasni, Doost Ali Mojdeh . Maximal first Zagreb index of trees with given Roman domination number. AIMS Mathematics, 2022, 7(7): 11801-11812. doi: 10.3934/math.2022658 |
[5] | Bana Al Subaiei, Ahlam AlMulhim, Abolape Deborah Akwu . Vertex-edge perfect Roman domination number. AIMS Mathematics, 2023, 8(9): 21472-21483. doi: 10.3934/math.20231094 |
[6] | Rangel Hernández-Ortiz, Luis Pedro Montejano, Juan Alberto Rodríguez-Velázquez . Weak Roman domination in rooted product graphs. AIMS Mathematics, 2021, 6(4): 3641-3653. doi: 10.3934/math.2021217 |
[7] | Zepeng Li . A note on the bounds of Roman domination numbers. AIMS Mathematics, 2021, 6(4): 3940-3946. doi: 10.3934/math.2021234 |
[8] | Mingyu Zhang, Junxia Zhang . On Roman balanced domination of graphs. AIMS Mathematics, 2024, 9(12): 36001-36011. doi: 10.3934/math.20241707 |
[9] | Jian Yang, Yuefen Chen, Zhiqiang Li . Some sufficient conditions for a tree to have its weak Roman domination number be equal to its domination number plus 1. AIMS Mathematics, 2023, 8(8): 17702-17718. doi: 10.3934/math.2023904 |
[10] | Chang-Xu Zhang, Fu-Tao Hu, Shu-Cheng Yang . On the (total) Roman domination in Latin square graphs. AIMS Mathematics, 2024, 9(1): 594-606. doi: 10.3934/math.2024031 |
A signed double Italian dominating function (SDIDF) on a graph G=(V,E) is a function f from V to {−1,1,2,3}, satisfying (ⅰ) ∑u∈N[v]f(u)≥1 for all v∈V; (ⅱ) if f(v)=−1 for some v∈V, then there exists A⊆N(v) such that ∑u∈Af(u)≥3; and (ⅲ) if f(v)=1 for some v∈V, then there exists A⊆N(v) such that ∑u∈Af(u)≥2. The weight of an SDIDF f is ∑v∈Vf(v). The signed double Italian domination number of G is the minimum weight of an SDIDF on G. In this paper, we initiated the study of signed double Italian domination and proved that the decision problem associated with the signed double Italian domination is NP-complete. We also provided tight lower and upper bounds of a signed double Italian domination number of trees, and we characterized the trees achieving these bounds. Finally, we determined the signed double Italian domination number for some well-known graphs.
All graphs G=(V,E) in this paper are finite, simple, and undirected. Two vertices, u,v∈V, are adjacent if uv∈E. We say u is a neighbor of v if u and v are adjacent. We denote the set of all neighbors of v by N(v) and the set N(v)∪{v} by N[v]. The degree of a vertex v is d(v):=|N(v)|. The maximum degree of G is Δ(G):=max{d(v)|v∈V} and the minimum degree of G is δ(G):=min{d(v)|v∈V}. The notation Pn denotes a path of order n, and the length of a path is the number of edges in it. The distance between two vertices u and v in a connected graph G, denoted by distG(u,v), is the length of a shortest path between u and v in G. The graph diameter is defined as diam(G):=max{distG(x,y)|(x,y)∈V×V}. A vertex u with d(u)=1 is called a leaf, and any neighbor of a leaf is called a support vertex. We denote the set of leaves in G by L(G) and the set of support vertices in G by S(G). A vertex v is called a strong support vertex if v∈S(G) and |N(v)∩L(G)|≥2, and v is called a weak support vertex if v∈S(G) and |N(v)∩L(G)|=1. An edge e∈E is a pendant edge if it has an endpoint in L(G). A tree with exactly two non-leaf vertices, u and v, is called a double star and it is denoted by DSs,t, where s=d(u) and t=d(v). We denote the set of integers {1,⋯,n} by [n]. If k is a positive integer less than n, then [n]∖[k]:={k+1,⋯,n}. A decision problem is an NP problem if it can be solved in polynomial time by a non-deterministic machine.
Let G be a graph and let f be a function f:V⟶A, where A is a finite subset of Z. The weight of f, denoted by w(f), is the sum ∑v∈Vf(v). If H⊆V, we denote the sum ∑v∈Hf(v) by f(H). Every function f can be represented by the partition (Vfi|i∈A) of V, where Vi={v∈V|f(v)=i}. Sometimes, we write Vi instead of Vfi if f is known from the context.
A function f from V to {0,1,2} is a Roman dominating function on G if every vertex in Vf0 is adjacent to at least one vertex in Vf2. The Roman domination number of G, denoted by γR(G), is the minimum weight of a Roman dominating function on G.
Roman domination was sparked by defensive actions taken to safeguard the Roman Empire. Every city was to have a maximum of two legions stationed there, according to a decree made by Constantine the Great, who was born in 272 and died in 337 BCE. Additionally, if a city does not have a legion, it must be near a city with two legions; in the event that the city without a legion was under attack, the city with two legions may dispatch a legion to defend it. Roman dominion was considered a mathematical notion by Stewart [1], ReVelle and Rosing [2,3] and was later developed by Cockayne et al. [4]. Clearly, the defense strategy would be insufficient to safeguard the empire if two attacks occurred simultaneously. This increased the necessity for a more effective defensive strategy and inspired scholars to develop and examine several Roman domination variations. Since then, more than a hundred studies on Roman dominance and its variations have been published. Italian domination [5], double Roman domination [6], signed Roman domination [7] and signed Italian domination [8] are a few examples of Roman domination variants. We refer the readers to [9,10,11,12,13,14] for recent work in Roman domination variants.
Chellali et al. [5] relaxed the conditions of Roman domination and introduced Italian domination. A function f:V⟶{0,1,2} is an Italian dominating function on G, denoted by IDF, if every vertex in Vf0 is adjacent to at least one vertex in Vf2 or at least two vertices in Vf1. The Italian domination number of G is γI(G):=min{w(f)| f is an IDF on G}.
A function f:V⟶{−1,1,2,3} is a signed double Roman dominating function on G, denoted by SDRDF, if (ⅰ) for all v∈V, ∑z∈N[v]f(z)≥1; (ⅱ) every vertex in Vf−1 is adjacent to at least one vertex in Vf3 or adjacent to at least two vertices in Vf2 and (ⅲ) every vertex in Vf1 is adjacent to at least one vertex in Vf2∪Vf3. The signed double Roman domination number of G, denoted by γsdR(G), is the minimum weight of an SDRDF on G. Signed double Roman domination was first introduced in [15]. In a signed double Roman domination model, a vertex with label negative one represents a location with an auxiliary cohort. The idea of this model comes from the era of Constantine Ⅱ, when small sizes of non-Roman citizen troops were formed (the size was about 10 of a legion). Constantine Ⅱ reduced the number of legions by stationing those troops in some locations and requiring that the number of troops in a location and its surrounding was less than the number of legions.
In this paper, we relax the conditions of signed double Roman domination and introduce signed double Italian domination (SDID). Our motivation is to reduce the cost while still protecting the empire.
Definition 1. Let G=(V,E) be a graph. A function f:V⟶{−1,1,2,3} is called a signed double Italian dominating function (SDIDF) on G if (i) ∑u∈N[v]f(u)≥1 for every v∈V; (ii) for every vertex v with f(v)=−1, there exists A⊆N(v) such that ∑u∈Af(u)≥3 and (iii) for every vertex v with f(v)=1, there exists A⊆N(v) such that ∑u∈Af(u)≥2. The signed double Italian domination number of G, denoted by γsdI(G), is the minimum weight of a SDIDF on G. A SDIDF on G with minimum weight is called a γsdI(G)-function.
It is obvious that every SDRDF on G is a SDIDF, so we have the following proposition.
Proposition 1. Let G be a graph, then γsdI(G)≤γsdR(G).
In the following proposition, we give examples of trees T for which γsdI(T)≠γsdR(T).
Proposition 2. There exist infinitely many trees T for which γsdI(T)<γsdR(T).
Proof. Let k be a positive integer. Let T4k be the tree obtained from the star graph K1,4k by subdividing every edge once. Denote the vertex in the center by c. Let the set of vertices with degree two be S:={v1,v2,⋯v4k}, and let the set of leaves be L:={u1,u2,⋯u4k}. Fix the notations so that vlul∈E(T4k). We show that γsdI(T4k)≤5k+1, while γsdR(T4k)>5k+1. Define an SDIDF f on T4k as follows. Set f(c)=1, set f(vl)=−1 and f(ul)=2 for every l∈[3k], and set f(vl)=3 and f(ul)=−1 for every l∈[4k]∖[3k]. As f is an SDIDF with w(f)=5k+1, then γsdI(T4k)≤5k+1.
Assume that T4k admits an SDRDF g with w(g)≤5k+1. Observe that for every l∈[4k], ∑v∈N[ul]g(v)=g(ul)+g(vl)≥1. As w(g)≤5k+1, there exists l∈[4k] such that g(ul)+g(vl)=1. We must have g(ul)=2 and g(vl)=−1, so g(c)∈{2,3}. Let a:=|{l∈[4k]|g(ul)+g(vl)>1}|, so |{l∈[4k]|g(ul)+g(vl)=1}|=4k−a. If a≥k, then
w(g)≥2a+4k−a+2=a+4k+2≥k+4k+2>5k+1; |
a contradiction. Thus, a≤k−1. Now,
g(N[c])≤3a−(4k−a)+g(c)≤4a−4k+3≤4(k−1)−4k+3≤−1; |
a contradiction. Thus, γsdR(T4k)>5k+1 as desired.
Proposition 3. Let G be a connected graph of order n≥2, then γsdI(G)≤n.
Proof. If n=2, then G=P2 and the statement is clear. Assume that n>2. If δ(G)≥2, we can assign value one for all v∈V. This gives an SDIDF f on G with w(f)=n. So, we may assume that δ(G)=1. Define an SDIDF g on G as follows. Assign value one to every vertex v∈V(G)∖(S(G)∪L(G)). For every v∈S(G), assign value three to v, assign value negative one to one of its leaves, and assign value one to the remaining leaves. Clearly, w(g)=n, and, thus, γsdI(G)≤n.
Let G be a connected graph of order n≥2. Let
A={v|v be a strong support vertex in G},
B={v∈A||N(v)∩L(G)| be odd},
C={v|v be a leaf adjacent to a strong support vertex}.
Proposition 4. Let G be a connected graph of order n>2 and G≠K1,n−1, then γsdI(G)≤n−(|B|+|C|).
Proof. Define an SDIDF f on G as follows. Set f(v)=1 for all v∈V(G)∖(S(G)∪L(G)). Let v∈S(G) and N(v)∩L(G)={v1,⋯,vk}, k≥1. Set f(v)=3, f(v1)=−1; if they exist, set f(v2)=f(v3)=−1, set f(vj)=1 if j is even and j≠2 and set f(vj)=−1 if j is odd. Do this for every support vertex v. As G≠K1,n−1, every v∈S(G) has a neighbor in V(G)∖L(G) and every non-leaf vertex is assigned a positive value; thus, f(N[v])≥1. The other conditions of the SDIDF are clearly satisfied. Therefore, γsdI(G)≤w(f)≤n−(|B|+|C|) as desired.
In this section, we prove that the decision problem associated with SDID is NP-complete for bipartite and chordal graphs.
Define the following problem.
SDID
Instance: Graph G and an integer k≤|V|.
Question: Does G have an SDIDF of weight at most k?
Our reduction is from the following known NP-complete problem.
EXACT COVER BY 3-SETS-3 (X3C3)
Instance: A set X of size 3q and a collection C={C1,⋯,C3q} of three-element subsets of X such that every element xi∈X is in exactly three elements of C.
Question: Does C contain an exact cover of X, i.e., does C contain a sub-collection C′ such that every element xi∈X is in exactly one element of C′?
Hickey et al. [16] proved that X3C3 is NP-complete.
We then transform every instance of X3C3 to an instance of SDID, such that the first instance has a solution if, and only if, the second instance has a solution.
Theorem 1. SDID is NP-complete for bipartite graphs and chordal graphs.
Proof. SDID is a member of NP, as we can check in polynomial time if a function f:V⟶{−1,1,2,3} is an SDIDF and w(f)≤k. Let X={x1,⋯,x3q} and C={C1,⋯,C3q} be an instance of X3C3. We construct an instance of SDID. For every i∈[3q], let xi be the center of the star graph K1,6. For every j∈[3q], let aj be the center of the star graph K1,4, and let cj be one of its leaves. Finally, add the set of edges xicj if, and only if, xi∈Cj. Call the resulting graph G1. Let G2 be the graph obtained from G1 by adding the edges cicj, where i,j∈[3q] and i≠j (see Figure 1). It is clear that G1 is a bipartite graph and G2 is a chordal graph. Let k=−5q.
Let G∈{G1,G2}. Assume that (X,C) has an exact cover C′. Define a function f on G as follows. Let f(v)=−1 for every leaf v in G, let f(xi)=f(ai)=3 for all i∈[3q] and let f(cj)=1 if Cj∉C′ and f(cj)=2 if Cj∈C′. It is clear to see that f is an SDIDF of weight equal to k.
Conversely, assume that G admits an SDIDF f with w(f)≤k. Choose a γsdI(G)-function f such that |Vf−1∩L(G)| is maximum possible. So, f(ai)=f(xi)=3 for every i∈[3q] and f(v)=−1 for every v∈L(G). As ∑u∈N[ai]f(u)≥1, then f(ci)≥1 for every i∈[3q]. As ∑z∈N[xi]f(z)≥1, then ∑z∈N(xi)∖L(G)f(z)≥4 for every i∈[3q]. Thus, ∑i∈[3q]f(ci)≥4(3q)3=4q. As w(f)≤−5q, then we must have ∑i∈[3q]f(ci)=4q. So, ∑z∈N(xi)∖L(G)f(z)=4 for every i∈[3q]. Thus, every xi is a neighbor of only one cj, with f(cj)=2. Hence, C′:={Cj∈C|f(cj)=2} is an exact cover.
In this section, we determine the tight lower and upper bounds of γsdI(T), where T is a tree, and we characterize trees achieving these bounds.
We start by determining γsdI(G) of the paths and star graphs, and we give a tight lower bound for double star graphs, which are simple examples of trees.
Proposition 5. Let n≥1, then
γsdI(Pn)={n3,ifn≡0mod3,⌊n3⌋+2,ifn≡1,2mod3. |
Proof. Let An=n3 if n≡0mod3 and An=⌊n3⌋+2 if n≡1,2mod3. Let Pn=v1v2⋯vn. First, we show that γsdI(Pn)≤An. Define an SDIDF f on Pn as follows. If n≡0mod3, let f(v3i−1)=3 for every i∈[n3] and let f=−1 otherwise. If n≡1mod3, let f(v3i−1)=3 for every i∈[⌊n3⌋], let f(vn)=2 and let f=−1 otherwise. If n≡2mod3, let f(v3i−1)=3 for every i∈[⌊n3⌋], let f(vn)=3 and let f=−1 otherwise. Thus, γsdI(Pn)≤w(f)=An.
Now, we show that γsdI(Pn)=An. Let f be a γsdI(Pn)-function. Assume that n≡0mod3, then
w(f)=∑i∈[n3]f(N[v3i−1])≥∑i∈[n3]1=n3. |
Thus, we must have γsdI(Pn)=w(f)=n3 as desired.
Assume that n≡1mod3. If f(vn)≥2 (this case is symmetric to f(v1)≥2), then
w(f)=∑i∈[⌊n3⌋]f(N[v3i−1])+f(vn)≥∑i∈[⌊n3⌋]1+2=⌊n3⌋+2, |
as desired. Assume that f(vn)<2. If f(vn)=1, then f(vn−1)=2, so it is not possible to have f(vn−2)=f(vn−3)=−1. Thus,
w(f)=∑i∈[⌊n3⌋−1]f(N[v3i−1])+f(N[vn−2])+f(vn)≥(⌊n3⌋−1)+2+1=⌊n3⌋+2, |
as desired. If f(vn)=−1, then f(vn−1)=3. Due to the symmetry of Pn, we can assume f(v1)=−1 and f(v2)=3, then
w(f)=f(v1)+f(v2)+f(vn−1)+f(vn)+∑i∈[⌊n3⌋−1]f(N[v3i+1])≥4+∑i∈[⌊n3⌋−1]1=⌊n3⌋+3; |
a contradiction. Thus, γsdI(Pn)=w(f)=An.
Assume that n≡2mod3. If f(vn−1)+f(vn)≥2, then
w(f)=∑i∈[⌊n3⌋]f(N[v3i−1])+f(vn−1)+f(vn)≥∑i∈[⌊n3⌋]1+2=⌊n3⌋+2, |
as desired. If f(vn−1)+f(vn)≤1, we must have f(vn)=2 and f(vn−1)=−1. Due to the symmetry of Pn, we may assume that f(v1)=2 and f(v2)=−1. Observe that f(v3)≥1, so f(N[v2])≥2, then
w(f)=∑i∈[⌊n3⌋]f(N[v3i−1])+f(vn−1)+f(vn)≥⌊n3⌋+2, |
as desired. Thus, γsdI(Pn)=w(f)=An.
Now, we give the exact SDID number of star graphs.
Proposition 6. Let n≥1, then
γsdI(K1,n)={2,ifn=1,3,1,otherwise. |
Proof. Denote the vertex in the center by c and denote the leaves by v1,⋯,vn. Observe that γsdI(K1,n)=f(N[c])≥1, where f is any SDIDF on K1,n. From Proposition 5, we get γsdI(K1,1)=2 and γsdI(K1,2)=1. Assume that n=3. By assigning value three to c, value negative one to each of v1 and v2 and value one to v3, we get an SDIDF of weight equal to two, so γsdI(K1,3)≤2. Clearly, γsdI(K1,3)=2. Assume that n≥4. Assign value three to c. If n is even, assign value negative one to each of v1 and v2. For the remaining vertices, assign value one to vi if i is even, and assign value negative one to vi if i is odd. If n is odd, assign value two to v1 and assign value negative one to each of v2,v3,v4,v5. For the remaining vertices, assign value one to vi if i is even and assign value negative one to vi if i is odd. So, γsdI(K1,n)≤1, and, thus, γsdI(K1,n)=1.
In Proposition 7, a tight lower bound of the SDID number of the double star graphs is given.
Proposition 7. Let s≥t≥1, then
(1) γsdI(DSs,t)≥−4 and
(2) γsdI(DSs,t)≥−5(t+s+2)+249, with equality if, and only if, s=t=5.
Proof. Denote the non-leaf vertex with degree s by u, the leaves adjacent to u by u1,⋯,us; the non-leaf vertex with degree t by v and the leaves adjacent to v by v1,⋯,vt. Let f be a γsdI(DSs,t)-function. Observe that γsdI(DSs,t)=f(N[u])+f(N[v])−f(u)−f(v)≥1+1−3−3=−4. Thus, the first statement holds.
If f(u)≤2 (this case is symmetric to f(v)≤2), then f(ui)≥1 for every i∈[s]. Thus,
γsdI(DSs,t)≥f(N[v])+s≥1+s>1>−5(s+t+2)+249, |
as desired. So, we may assume that f(u)=f(v)=3. If s≤5, then ∑i=si=1f(ui)=−s; similarly, if t≤5, then ∑i=ti=1f(vi)=−t. If s=6, then ∑i=si=1f(ui)=−4; similarly if t=6, then ∑i=ti=1f(vi)=−4. If s≥7, then ∑i=si=1f(ui)=−5; similarly if t≥7, then ∑i=ti=1f(vi)=−5.
If t,s≤5 then
w(f)=6−s−t≥−5(s+t+2)+249, |
with equality if, and only if, s=t=5. If s=6 and t≤5, then
w(f)=2−t>−5(s+t+2)+249. |
If s=t=6, then
w(f)=−2>−5(s+t+2)+249. |
If s≥7 and t≤5, then
w(f)=1−t>−5(s+t+2)+249. |
If s≥7 and t=6, then
w(f)=−3>−5(s+t+2)+249. |
If t,s≥7, then
w(f)=−4>−5(s+t+2)+249. |
Thus, the second statement holds.
Let C be the set of all graphs that can be obtained from a tree T by adding 3dT(v)+2 pendant edges to v for all v∈T. Next, we give the tight lower bound of the SDID number of the trees and characterize the trees achieving this bound.
Theorem 2. Let T be a tree of order n≥2, then
γsdI(T)≥−5n+249, |
with equality if, and only if, T∈C.
Proof. We use induction on the number of vertices. If n=2, then T=P2 and γsdI(P2)=2>−5n+249. When n=3, T=P3 and γsdI(P3)=1=−5n+249. Observe that P3∈C. This establishes the base step. Assume that n≥4 and that the statement holds for every tree of order less than n. Let f be a γsdI(T)-function, where T is a tree of order n. If diam(T)=2, then T=K1,n−1, so γsdI(T)≥1>−5n+249 (Proposition 6). If diam(T)=3, then T=DSs,t and γsdI(T)≥−5n+249, with equality if, and only if, T=DS5,5 (Proposition 7). Observe that DS5,5∈C. So, we may assume that diam(T)≥4.
If |Vf−1|≤1, then
w(f)≥n−2>−5n+249, |
and we are done. So, we may assume that |Vf−1|≥2. Assume that there exist two adjacent vertices u,v∈Vf−1. Let T1 and T2 be the trees obtained from T by deleting the edge uv. Clearly, |T1|,|T2|≥2, and the restriction of f on Ti is an SDIDF on Ti. From the induction hypothesis,
w(f)=f(T1)+f(T2)≥−5|T1|+249+−5|T2|+249=−5n+489>−5n+249, |
and we are done. So, we may assume that for every two vertices u,v∈Vf−1, distT(u,v)≥2. The argument is divided into the following cases.
Case 1. There exists x∈Vf−1 with d(x)≥2, then x is not a leaf. Denote the leaves adjacent to x, if any, by u1,⋯,us, and denote the subtrees of order greater than one obtained from T by deleting x by T1,⋯,Tr. It is clear that fi:=f|Ti is an SDIDF on Ti. Observe that r≥1 as diam(T)≥4 and f(ui)≥2 for all i∈[s] (if uis exist). Observe also that if r=1, then s≠0, as d(x)≥2. From the induction hypothesis,
w(f)≥∑i∈[r]−5|Ti|+249+∑i∈[s]f(ui)+f(x)≥−5(n−s−1)+24r9+2s−1=−5n+24r+23s−49>−5n+249, |
and we are done. So, we may assume that d(x)=1 for every x∈Vf−1. In other words, every vertex in Vf−1 is a leaf.
Case 2. There exists x∈Vf1 with d(x)≥2. So, Vf1 has a non-leaf vertex x. Denote the leaves adjacent to x, if any, by u1,⋯,us. So, f(ui)≥2 for every i∈[s] (if uis exist). Denote the subtrees of order greater than one that are obtained from T by deleting x by T1,⋯,Tr. Observe that r≥1 as diam(T)≥4, and if s=0, then r≥2. Denote the unique vertex adjacent to x in Ti by yi. Observe that f(yi)≥1, as yi is a non-leaf vertex. For every i∈[r], do the following. If f(yi)=1 or 2, define an SDIDF g on Ti as g(yi)=f(yi)+1 and g(a)=f(a) for every a∈Ti∖{yi}. If f(yi)=3 and f(N[yi])≥2, then fi:=f|Ti is an SDIDF on Ti, and if f(N[yi])=1 then, yi is adjacent to a vertex w∈Vf−1. Define an SDIDF g on Ti, as g(w)=1 and g(a)=f(a) for every a∈Ti∖{w}. From the induction hypothesis,
γsdI(T)=w(f)≥∑i∈[r]f(Ti)+2s+1≥∑i∈[r]−5|Ti|+249−2r+2s+1=−5(n−s−1)+24r9−2r+2s+1=−5n+6r+23s+149>−5n+249. |
The last inequality follows from the fact that r≥1, and if r=1, then s≥1. So, we may assume that every non-leaf vertex is assigned two or three.
Case 3. There exists x∈Vf2 with d(x)≥2; in other words, x is not a leaf. Denote the leaves adjacent to x, if any, by u1,⋯,us. So, f(ui)≥1 for every i∈[s] (if uis exist). Denote the subtrees of order greater than one that are obtained from T by deleting x by T1,⋯,Tr. Observe that r≥1 as diam(T)≥4. As in the previous case, denote the unique vertex adjacent to x in Ti by yi. Observe that f(yi)≥2, as yi is a non-leaf vertex. For every i∈[r], do the following. If f(yi)=2, define an SDIDF g on Ti as g(yi)=3 and g(a)=f(a) for every a∈Ti∖{yi}. If f(yi)=3 and f(N[yi])≥3, then fi:=f|Ti is an SDIDF on Ti, and if f(N[yi])∈{1,2}, then yi is adjacent to a vertex w∈Vf−1. Define a SDIDF g on Ti, as g(w)=1 and g(a)=f(a) for every a∈Ti∖{w}. From the induction hypothesis,
γsdI(T)=w(f)≥∑i∈[r]f(Ti)+s+2≥∑i∈[r]−5|Ti|+249−2r+s+2=−5(n−s−1)+24r9−2r+s+2=−5n+6r+14s+239>−5n+249. |
The last inequality follows from the fact that r≥1.
So, we may assume that all non-leaf vertices are labeled three.
Case 4. There exists x∈T such that x is neither a leaf nor a support vertex. Let N(x)={y1,⋯,yr}, r≥2. Observe that f(x)=f(yi)=3 for all i∈[r]. Let T′ be the tree obtained from T be deleting x and adding the set of edges {yiyi+1|i∈[r−1]}. Let g be an SDIDF on T′ defined as g(u)=f(u) for all u∈T′. From the induction hypothesis,
γsdI(T)=w(g)+f(x)≥−5(n−1)+249+3=−5n+569>−5n+249, |
as required. So, assume L(T)∪S(T)=V(T).
Assume that there exists a leaf y such that f(y)≥1. Denote the support vertex adjacent to y by x. Recall that f(x)=3. Since f is a γsdI(G)-function, there are three leaves zi,i∈[3] in N(x) such that f(zi)=−1 for all i∈[3]. Let T′=T−{z1,y} if f(y)=1, let T′=T−{z1,z2,y} if f(y)=2 and let T′=T−{z1,z2,z3,y} if f(y)=3. Define an SDIDF g on T′ by setting g(v)=f(v) for all v∈T′. Then,
γsdI(T)=w(g)≥−5|T′|+249≥−5(n−2)+249>−5n+249, |
so the statement holds.
So, we may assume that all leaves are labeled negative one.
Let T′=T−Vf−1. It can be seen that for every support vertex x∈T, the number of leaves adjacent to x in T is at most 3dT′(x)+2. It is well known that for any graph G, ∑v∈Gd(v)=2|E(G)|. It is also well known that for any tree T, |E(T)|=|T|−1. Thus,
|Vf−1|≤∑x∈T′(3dT′(x)+2)=3(2(|T′|−1))+2|T′|=8|T′|−6. |
So,
4|Vf−1|≤32|T′|−24. | (3.1) |
Now,
γsdI(T)=3|T′|−|Vf−1|=27|T′|−9|Vf−1|9≥(∗)−5|T′|−5|Vf−1|+249=−5(|T′|+|Vf−1|)+249=−5n+249, |
where inequality (∗) follows from Eq (3.1). Observe that the equality holds if |Vf−1|=8|T′|−6, i.e., every x∈T′ is adjacent to 3dT′(x)+2 leaves.
Conversely, assume that T∈C. Let h be an SDIDF on T such that h(v)=−1 if v is a leaf and h(v)=3 if v is a support vertex. Then, γsdI(T)≤w(h)=−5n+249, and, thus, γsdI(T)=−5n+249.
Proposition 8. Let T be a tree with |T|≥2, then
(1) γsdI(T)≤|T| and
(2) γsdI(T)=|T| if, and only if, T=P2.
Proof. It follows directly from Proposition 1 and the fact that γsdR(T)≤|T| [15].
Proposition 9. Let G be a connected graph with |G|≥3. If γsdI(G)=|G|, then δ(G)≥2.
Proof. Assume for contrast that δ(G)=1. From Proposition 6, G≠K1,|G|−1. From Proposition 4, G does not have a strong support vertex. So, all the support vertices are weak. Let v be one of the weak support vertices. Denote the leaf adjacent to v by u. If v is adjacent to a vertex w such that d(w)=2 and w is not a support vertex, assign value negative one to u and w, value three to v, value two to the other neighbor of w, value three to each of the remaining support vertices, value negative one to each of the remaining leaves, and value one to the remaining vertices. If all the non-leaf neighbors of v are support vertices or have a degree greater than two, assign value two to u, assign value negative to v, assign value three to each of the remaining support vertices, assign value negative one to each of the remaining leaves, and assign value one to remaining vertices. Clearly this labeling is an SDIDF of weight less than |G|, so γsdI(G)<|G|; a contradiction. Thus, δ(G)≥2.
In this section, γsdI(G) is determined for cycle graphs and the Petersen graph.
Proposition 10. Let Cn be a cycle graph of order n, then
γsdI(Cn)={n3,ifn≡0mod3,n−43+4,ifn≡1mod3,n−23+2,ifn≡2mod3. |
Proof. Let Cn=v1v2⋯vn. Assume that n is a multiple of three. Assume that f is an SDIDF on Cn, then w(f)=∑i≡2mod3f(N[vi])≥∑i≡2mod31=n3. Thus, γsdI(Cn)≥n3. Now, we show that n3 is sufficient. Set g(vi)=3 if i≡2mod3 and set g(vi)=−1 otherwise, then g is an SDIDF on Cn with w(g)=n3. Thus, γsdI(Cn)≤n3. Therefore, γsdI(Cn)=n3 if n is a multiple of three.
Assume that n is not a multiple of three. We use induction on n. It is easy to check that γsdI(C4)=4 and γsdI(C5)=3. Assume that there exists an SDIDF f with w(f)≤an−1, where an=n−43+4 if n≡1mod3 and an=n−23+2 if n≡2mod3.
Claim 1. an−2<an−3.
Proof. If an=n−43+4, then
an−2=n−43+2=(n−3)−43+3<an−3. |
If an=n−23+2, then
an−2=n−23=(n−3)−23+1<an−3. |
Case 1. dist(x,y)=1.
Say that x=vi and y=vi+1. Observe that vi−1 and vi+2 each are assigned value of three. Contract the path vi−1vivi+1vi+2 into a vertex v. The resulting graph is Cn−3. Define an SDIDF g on Cn−3 by setting g(v)=3 and g(u)=f(u) for all u∈V(Cn−3)∖{v}, then w(g)=w(f)−1≤an−2<an−3, contradicting the induction hypothesis. So, the statement holds.
Case 2. dist(x,y)=2.
Say that x=vi and y=vi+2. We must have f(vi+1)=3 and f(vi+3)≥1. Contract the path vivi+1vi+2vi+3 into a vertex v. The resulting graph is the cycle Cn−3. Define an SDIDF g on Cn−3 as follows. Set g(v)=f(vi+3) and g(u)=f(u) for all u∈V(Cn−3)∖{v}, then, w(g)≤w(f)−1≤an−2<an−3, contradicting the induction hypothesis. So, the statement holds.
Case 3. dist(x,y)≥3.
Say that x=vi, then f(vi−2),f(vi−1),f(vi+1),f(vi+2)≥1. Contract the path vi−1vivi+1vi+2 into a vertex v. The resulting graph is the cycle Cn−3. Define an SDIDF g on Cn−3 as follows. Set g(v)=max{f(vi−1),f(vi+1),f(vi+2)} and g(u)=f(u) for all u∈V(Cn−3)∖{v}, then w(g)≤w(f)−1≤an−2<an−3, contradicting the induction hypothesis. So, the statement holds.
Proposition 11. The SDID number for the Petersen graph is five.
Proof. Let G be the Petersen graph with vertices labeled as in Figure 2. Define an SDIDF f on G as follows. Set f(x1)=f(x4)=f(y2)=f(y3)=−1, set f(x2)=f(y4)=f(y5)=1 and set f(x3)=f(x5)=f(y1)=2, then γsdI(G)≤w(f)=5. To show that five is tight, assume that G admits an SDIDF g of weight less than five, then |Vg−1|≥3. We divide the argument into two cases.
Case 1. There are u,v∈Vg−1 with uv∈E.
Due to the symmetry of the Petersen graph, assume that g(x1)=g(x2)=−1, then g(x5)+g(y1)≥3 and, similarly, g(x3)+g(y2)≥3. Thus, at least two vertices from the remaining vertices, i.e., x4,y3,y4,y5, are assigned value negative one. Note that it is impossible to have {y3,y4}⊆Vg−1, as we would have g(N[y1])≤0; similarly, it is impossible to have {y4,y5}⊆Vg−1, as we would have g(N[y2])≤0. It is also impossible to have {x4,y5}⊆Vg−1, as we would have g(N[x5])≤0, and it is impossible to have {x4,y3}⊆Vg−1, as we would have g(N[x3])≤0. Thus, we are left with two sub-cases. The first sub-case is {x4,y4}⊆Vg−1 and {y3,y5}⊆Vg1, and the second sub-case is {y3,y5}⊆Vg−1 and {x4,y4}⊆Vg1. If we are in the first sub-case, we must have g(x3),g(x5),g(y1),g(y2)≥2. Thus, w(g)>4, a contradiction. We get the same result if we are in the second sub-case. Thus, Case 1 is impossible.
Case 2. Vg−1 is an independent set.
Due to the symmetry of G, we may assume that g(x1)=g(x3)=−1. As each of x2,x4,x5,y1,y3 is adjacent to x1 or x3, then g(v)≥1 for all v∈{x2,x4,x5,y1,y2}. Note that g(y2)≥1 or else we would have g(N[x2])≤0, which is a contradiction. Thus, either g(y4)=−1 or g(y5)=−1. Due to the symmetry of G, we may assume that g(y4)=−1. As w(g)≤4, then y5 is assigned one or negative one. If g(y5)=1, then g(x2),g(x4),g(x5),g(y1),g(y2)=1, but now we have g(N[x2])=0; a contradiction. Thus, g(y5)=−1. Observe that it is impossible to have g(x2)=g(y2)=1, as we would have g(N[x2])=0, so g(x2)+g(y2)≥3. Similarly, it is impossible to have g(y1)=g(y3)=1, as we would have g(N[y1])=0, so g(y1)+g(y3)≥3. It is also impossible to have g(x4)=g(x5)=1, as we would have g(N[x5])=0, so g(x4)+(x5)≥3. Thus, w(g)≥3(3)−4=5; a contradiction. Hence, the statement holds.
We conclude this work with two open problems.
Problem 1. Characterize connected graphs G with |G|≥3, satisfying γsdI(G)=|G|.
Problem 2. Give characterizations for graphs G satisfying γsdI(G)=γsdR(G).
The author declares she has not used Artificial Intelligence (AI) tools in the creation of this article.
This work was supported by the Deanship of Scientific Research, Vice Presidency for Graduate Studies and Scientific Research, King Faisal University, Saudi Arabia [Grant No. 4988].
The author declares that she has no conflict of interest.
[1] | I. Stewart, Defend the Roman empire, Sci. Am., 281 (1999), 136–138. |
[2] | Johns Hopkins magazine, Can you protect the Roman empire? C. ReVelle, 1997. Available from: https://pages.jh.edu/jhumag/0497web/locate3.html. |
[3] | Johns Hopkins magazine, Test your solution to can you protect the Roman empire? C. ReVelle, 1997. Available from: https://pages.jh.edu/jhumag/0697web/revelle.html. |
[4] |
E. Cockayne, P. Dreyer, S. M. Hedetniemi, S. T. Hedetniemi, Roman domination in graphs, Discrete Math., 278 (2004), 11–22. http://dx.doi.org/10.1016/j.disc.2003.06.004 doi: 10.1016/j.disc.2003.06.004
![]() |
[5] |
M. Chellali, T. Haynes, S. Hedetniemi, A. McRae, Roman 2-domination, Discrete Appl. Math., 204 (2016), 22–28. http://dx.doi.org/10.1016/j.dam.2015.11.013 doi: 10.1016/j.dam.2015.11.013
![]() |
[6] |
R. Beeler, T. Haynes, S. Hedetniemi, Double Roman domination, Discrete Appl. Math., 211 (2016), 23–29. http://dx.doi.org/10.1016/j.dam.2016.03.017 doi: 10.1016/j.dam.2016.03.017
![]() |
[7] |
H. Abdollahzadeh Ahangar, M. Henning, C. Löwenstein, Y. Zhao, V. Samodivkin, Signed Roman domination in graphs, J. Comb. Optim., 27 (2014), 241–255. http://dx.doi.org/10.1007/s10878-012-9500-0 doi: 10.1007/s10878-012-9500-0
![]() |
[8] | A. Karamzadeh, H. Maimani, A. Zaeembashi, On the signed Italian domination of graphs, Comput. Sci. J. Mold., 27 (2019), 204–229. |
[9] |
T. Haynes, M. Henning, Perfect Italian domination in trees, Discrete Appl. Math., 260 (2019), 164–177. http://dx.doi.org/10.1016/j.dam.2019.01.038 doi: 10.1016/j.dam.2019.01.038
![]() |
[10] |
A. Egunjobi, T. Haynes, Perfect double Roman domination of trees, Discrete Appl. Math., 284 (2020), 71–85. http://dx.doi.org/10.1016/j.dam.2020.03.021 doi: 10.1016/j.dam.2020.03.021
![]() |
[11] |
A. Almulhim, Total perfect Roman domination, Symmetry, 15 (2023), 1676. http://dx.doi.org/10.3390/sym15091676 doi: 10.3390/sym15091676
![]() |
[12] |
M. Henning, W. Klostermeyer, G. MacGillivray, Perfect Roman domination in trees, Discrete Appl. Math., 236 (2018), 235–245. http://dx.doi.org/10.1016/j.dam.2017.10.027 doi: 10.1016/j.dam.2017.10.027
![]() |
[13] |
H. Naresh Kumar, Y. Venkatakrishnan, Trees with vertex-edge Roman domination number twice the domination number minus one, Proyecciones, 39 (2020), 1381–1392. http://dx.doi.org/10.22199/issn.0717-6279-2020-06-0084 doi: 10.22199/issn.0717-6279-2020-06-0084
![]() |
[14] |
B. AlSubaiei, A. Almulhim, A. Akwu, Vertex-edge perfect Roman domination number, AIMS Mathematics, 8 (2023), 21472–21483. http://dx.doi.org/10.3934/math.20231094 doi: 10.3934/math.20231094
![]() |
[15] |
H. Ahangar, M. Chellali, S. Sheikholeslami, Signed double Roman domination in graphs, Discrete Appl. Math., 257 (2019), 1–11. http://dx.doi.org/10.1016/j.dam.2018.09.009 doi: 10.1016/j.dam.2018.09.009
![]() |
[16] |
G. Hickey, F. Dehne, A. Rau-Chaplin, C. Blouin, SPR distance computation for unrooted trees, Evol. Bioinform., 4 (2008), 17–27. http://dx.doi.org/10.4137/EBO.S419 doi: 10.4137/EBO.S419
![]() |
1. | Ahlam Almulhim, Bana Al Subaiei, Saiful Rahman Mondal, Survey on Roman {2}-Domination, 2024, 12, 2227-7390, 2771, 10.3390/math12172771 |