
Let f:V(G)→{0,1,2} be a function defined from a connected graph G. Let Wi={x∈V(G):f(x)=i} for every i∈{0,1,2}. The function f is called a total Italian dominating function on G if ∑v∈N(x)f(v)≥2 for every vertex x∈W0 and if ∑v∈N(x)f(v)≥1 for every vertex x∈W1∪W2. The total Italian domination number of G, denoted by γtI(G), is the minimum weight ω(f)=∑x∈V(G)f(x) among all total Italian dominating functions f on G. In this paper, we provide new lower and upper bounds on the total Italian domination number of trees. In particular, we show that if T is a tree of order n(T)≥2, then the following inequality chains are satisfied.
(ⅰ) 2γ(T)≤γtI(T)≤n(T)−γ(T)+s(T),
(ⅱ) n(T)+γ(T)+s(T)−l(T)+12≤γtI(T)≤n(T)+γ(T)+l(T)2,
where γ(T), s(T) and l(T) represent the classical domination number, the number of support vertices and the number of leaves of T, respectively. The upper bounds are derived from results obtained for the double domination number of a tree.
Citation: Abel Cabrera-Martínez, Andrea Conchado Peiró, Juan Manuel Rueda-Vázquez. Further results on the total Italian domination number of trees[J]. AIMS Mathematics, 2023, 8(5): 10654-10664. doi: 10.3934/math.2023540
[1] | Ahlam Almulhim . Signed double Italian domination. AIMS Mathematics, 2023, 8(12): 30895-30909. doi: 10.3934/math.20231580 |
[2] | Linyu Li, Jun Yue, Xia Zhang . Double total domination number of Cartesian product of paths. AIMS Mathematics, 2023, 8(4): 9506-9519. doi: 10.3934/math.2023479 |
[3] | Ana Klobučar Barišić, Antoaneta Klobučar . Double total domination number in certain chemical graphs. AIMS Mathematics, 2022, 7(11): 19629-19640. doi: 10.3934/math.20221076 |
[4] | 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 |
[5] | Abel Cabrera-Martínez, Andrea Conchado Peiró . On the {2}-domination number of graphs. AIMS Mathematics, 2022, 7(6): 10731-10743. doi: 10.3934/math.2022599 |
[6] | Zhihong Xie, Guoliang Hao, S. M. Sheikholeslami, Shuting Zeng . On the Italian reinforcement number of a digraph. AIMS Mathematics, 2021, 6(6): 6490-6505. doi: 10.3934/math.2021382 |
[7] | 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 |
[8] | Zepeng Li . A note on the bounds of Roman domination numbers. AIMS Mathematics, 2021, 6(4): 3940-3946. doi: 10.3934/math.2021234 |
[9] | 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 |
[10] | 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 |
Let f:V(G)→{0,1,2} be a function defined from a connected graph G. Let Wi={x∈V(G):f(x)=i} for every i∈{0,1,2}. The function f is called a total Italian dominating function on G if ∑v∈N(x)f(v)≥2 for every vertex x∈W0 and if ∑v∈N(x)f(v)≥1 for every vertex x∈W1∪W2. The total Italian domination number of G, denoted by γtI(G), is the minimum weight ω(f)=∑x∈V(G)f(x) among all total Italian dominating functions f on G. In this paper, we provide new lower and upper bounds on the total Italian domination number of trees. In particular, we show that if T is a tree of order n(T)≥2, then the following inequality chains are satisfied.
(ⅰ) 2γ(T)≤γtI(T)≤n(T)−γ(T)+s(T),
(ⅱ) n(T)+γ(T)+s(T)−l(T)+12≤γtI(T)≤n(T)+γ(T)+l(T)2,
where γ(T), s(T) and l(T) represent the classical domination number, the number of support vertices and the number of leaves of T, respectively. The upper bounds are derived from results obtained for the double domination number of a tree.
In this article, we consider G as a simple graph of order n(G)=|V(G)| and size m=|E(G)|. Given a vertex v of G, N(v)={x∈V(G):xv∈E(G)} and N[v]=N(v)∪{v}. The degree of v in G, denoted by degG(v), is the cardinality of N(v). A vertex v∈V(G) is a leaf if degG(v)=1, and v is a support vertex if it is adjacent to a leaf. The set of leaves and support vertices are denoted by L(G) and S(G), respectively. The values l(G) and s(G) represent the number of leaves and the number of support vertices, respectively, i.e., l(G)=|L(G)| and s(G)=|S(G)|. A set of vertices D⊆V(G) is a dominating set of G if |N(x)∩D|≥1 for every vertex x∈V(G)∖D. The domination number of G, denoted by γ(G), is the minimum cardinality among all dominating sets of G. A γ(G)-set is a dominating set of G of cardinality γ(G).
For an arbitrary subset P of nonnegative reals, a function f:V(G)→P is a dominating function on G if the set {x∈V(G):f(x)>0} is a dominating set of G. In 1998, the authors of the books [1,2] exposed some of the first studied varieties of dominating functions in graphs. Similarly, in the last two decades, dominating functions have been topics of interest within domination theory in graphs. In particular, the study of the Roman dominating functions and their variants stands out.
Recently, a new variant of Roman domination, called total Italian domination number, was introduced in [3] and independently in [4], under the name of total Roman {2}-domination number. For a graph G with no isolated vertex, a total Italian dominating function (TIDF) on G is a dominating function f:V(G)→{0,1,2} which satisfies the following two conditions.
● Every vertex x∈V(G) for which f(x)=0 satisfies that ∑u∈N(x)f(u)≥2.
● The subgraph induced by the set {x∈V(G):f(x)≥1} has no isolated vertex.
Observe that the function f generates three sets W0, W1 and W2, where Wi={x∈V(G):f(x)=i} for i∈{0,1,2}. In such a sense, we write f(W0,W1,W2) so as to refer to the TIDF f. Sometimes we will introduce the notation f(V0,V1,V2), with vertex sets Vi instead of vertex sets Wi, in order to distinguish some functions. Given a set D⊆V(G), f(D)=∑x∈Df(x). The total Italian domination number of G, denoted by γtI(G), is the minimum weight ω(f)=∑x∈V(G)f(x)=|W1|+2|W2| among all TIDFs f(W0,W1,W2) on G. For simplicity, a TIDF f of weight ω(f)=γtI(G) will be called a γtI(G)-function.
The problem of computing the total Italian domination number of a graph is NP-hard [3,4]. This suggests obtaining closed formulas or giving tight bounds for this parameter. Further combinatorial results on total Italian domination can be found for example, in [5,6,7,8]. In [4,5] some results for the case of trees were presented. For instance, in [4] the authors showed that ⌈2(n(T)−l(T)+3)3⌉≤γtI(T)≤3n(T)+2s(T)4 for any tree of order n(T)≥4. Moreover, in [5] the authors characterized the trees T with γtI(T)=3γ(T).
In this paper we continue with the study of this parameter in trees. In such a sense, our main goal is to provide some new tight bounds on the total Italian domination number in trees. The article is organized as follows. In Section 2 we introduce some additional concepts and notation needed to develop the remaining sections. Sections 3 and 4 are devoted to obtaining new lower and upper bounds on the total Italian domination number in terms of order, domination number, number of support vertices and number of leaves of a tree. In particular, we show that if T is a tree of order n(T)≥2, then the following inequality chains hold.
(ⅰ) 2γ(T)≤γtI(T)≤n(T)−γ(T)+s(T).
(ⅱ) (n(T)+γ(T)+s(T)−l(T)+1)/2≤γtI(T)≤(n(T)+γ(T)+l(T))/2.
In addition, we show some classes of graphs for which the bounds above are achieved. We end with a concluding remark section, where we provide two interesting consequences derived from the new bounds given and propose some open problems.
We first present additional necessary terminology and notation. Given a graph G and a set D⊆V(G), N(D)=∪x∈DN(x) and N[D]=N(D)∪D, respectively. As usual, by G-D we denote the graph obtained from G by removing all the vertices of D and all the edges incident with a vertex in D. In addition, let Ss(G)={x∈S(G):|N(x)∩L(G)|≥2}. Moreover, a vertex x∈V(G) is a semi-support vertex if x∈N(S(G))∖(S(G)∪L(G)). The set of semi-support vertices is denoted by SS(G). The minimum and maximum degrees of a graph G will be denoted by δ(G) and Δ(G), respectively. For any two vertices x,y∈V(G), the distance d(x,y) between x and y is the minimum length of a x−y path. The diameter of G, denoted by diam(G), is the maximum distance among all pairs of vertices in G.
Given a rooted tree T (with root r) and a vertex v∈V(T)∖{r}, we say that a descendant of v is a vertex u∈V(T) such that the unique r−u path contains v. The set of descendants of v is denoted by D[v]. The maximal subtree at v, denoted by Tv, is the subtree of T induced by D[v].
A set D⊆V(G) is said to be a double dominating set (DDS) of G if |N[x]∩D|≥2 for every vertex x∈V(G). The double domination number of G, denoted by γ×2(G), is the minimum cardinality among all DDSs of G. A γ×2(G)-set is a DDS of G of cardinality γ×2(G). This parameter was introduced in [9] by Harary and Haynes, and has been extensively studied. For instance, we cite the recent works [10,11,12,13,14]. In addition, we observe that a set D⊆V(G) is a DDS of G if and only if there exists a TIDF f(W0,W1,W2) such that W1=D and W2=∅. Therefore, and by definition, it follows that γtI(G)≤γ×2(G).
Any other definitions that are of interest, will be introduced where needed.
The main goal of this section is to show that for any tree T of order n(T)≥2,
2γ(T)≤γtI(T)≤n(T)−γ(T)+s(T). | (3.1) |
The previous inequality chain will be deduced as a direct consequence of Theorem 3.2 and Corollary 3.5. Before, we shall need to introduce the following useful lemma.
Lemma 3.1. If G is a connected graph of order at least four, then there exists a γtI(G)-function g(W0,W1,W2) such that the following conditions hold.
(a) Ss(G)⊆W2 and |N(x)∩L(G)∩W0|≥|N(x)∩L(G)|−1 for every x∈Ss(G).
(b) V≤2(G)⊆W0∪W1, where V≤2(G)={x∈V(G):degG(x)≤2}.
Proof. Among all the γtI(G)-functions f(V0,V1,V2) satisfying that |Ss(G)∩V2| is maximum, let g(W0,W1,W2) be a function such that |V≤2(G)∩W2| is minimum.
We first suppose that there exists a vertex v∈Ss(G)∖W2. This implies that v∈W1 and N(v)∩L(G)⊆W1. Now, we consider the function g′(W′0,W′1,W′2) defined by g′(v)=2, g′(N(v)∩L(G))=g(N(v)∩L(G))−1 and g′(x)=g(x) whenever x∈V(G)∖({v}∪(N(v)∩L(G))). Since |N(v)∩L(G)|≥2, it follows that g′ is a γtI(G)-function such that |Ss(G)∩W′2|>|Ss(G)∩W2|, a contradiction. Therefore, Ss(G)⊆W2. In addition, and as an immediate consequence of the previous inclusion, we have that |N(x)∩L(G)∩W0|≥|N(x)∩L(G)|−1 for every x∈Ss(G). Hence, condition (a) follows.
Finally, we proceed to prove (b). Let x∈V≤2(G). If x∈L(G), then it is straightforward that x∈W0∪W1. Now, we assume that N(x)={y,z} and without loss of generality, suppose that g(y)≤g(z). If g(z)=0 then g(y)=0, and as a consequence, we have that N(x)⊆W0, which contradicts the fact that N(x)∩(W1∪W2)≠∅ by definition. Hence, g(z)>0. If x∈W2 then y∈W0, which implies that the function h(V′0,V′1,V′2) defined by h(x)=h(y)=1 and h(u)=g(u) whenever u∈V(G)∖{x,y}, is a γtI(G)-function such that |Ss(G)∩V′2|=|Ss(G)∩W2| (observe that x∉Ss(G) because |N(x)|=2 and n(G)≥4) and |V≤2(G)∩V′2|=|V≤2(G)∩(W2∖{x})|<|V≤2(G)∩W2|, a contradiction. Hence, x∈W0∪W1, which implies that V≤2(G)⊆W0∪W1. Therefore, the proof is complete.
In order to prove the next result, we need to introduce the following definition. A set S⊆V(G) is a 2-packing set of G if N[x]∩N[y]=∅ for every pair of different vertices x,y∈S. The 2-packing number of G, denoted by ρ(G), is the maximum cardinality among all 2-packing sets of G.
Theorem 3.2. For any connected graph G of order at least two,
γtI(G)≥γ(G)+s(G). |
Furthermore, for any tree T of order n(T)≥2,
γtI(T)≥2γ(T). |
Proof. If n(G)∈{2,3}, then it is straightforward that γtI(G)≥γ(G)+s(G). From now on, we assume that n(G)≥4. Let g(W0,W1,W2) be a γtI(G)-function defined as in the proof of Lemma 3.1. Hence, g satisfies the conditions given in Lemma 3.1. As S(G)⊆W1∪W2 and |Ss(G)∩W2| is maximum, then it is easy to deduce the following conditions.
(ⅰ) L(G)⊆W0∪W1.
(ⅱ) |N(x)∩L(G)∩W1|≤1 for every vertex x∈S(G).
By the previous conditions and the fact that S(G)⊆W1∪W2, we have that |S(G)∩W1|≤|L(G)∩W1| and that W2∪(W1∖L(G)) is a dominating set of G. Therefore,
γ(G)+s(G)=γ(G)+|S(G)|≤|W2∪(W1∖L(G))|+|S(G)|≤|W2|+|W1|−|L(G)∩W1|+|S(G)|≤|W2|+|W1|−|S(G)∩W1|+|S(G)|≤2|W2|+|W1|−|S(G)∩W1|−|S(G)∩W2|+|S(G)|=2|W2|+|W1|=γtI(G), |
as desired. Now, let T be any tree of order at least two and let S be a 2-packing set of T of cardinality ρ(T). From any γtI(T)-function f, it follows that f(N[x])≥2 for every x∈V(G). Since N[x]∩N[y]=∅ for every pair of different vertices x,y∈S, we deduce that
γtI(T)≥∑x∈Sf(N[x])≥2|S|=2ρ(T). |
Finally, the result follows due to the fact that γ(T)=ρ(T) for any tree T (see [15]).
Now, we consider the following family of trees. We say that a tree T belongs to the family T if it satisfies one of the following two conditions.
● T is a subdivided star, i.e., T is a graph obtained from a star by subdividing every edge exactly once.
● T can be obtained from a star K1,n by subdividing exactly n−1 edges at most twice.
The following theorem provides a lower bound for any tree T with s(T)=l(T). In addition, this result shows that the bounds given in Theorem 3.2 are achieved for any tree T belongs to the family T previously defined.
Theorem 3.3. The following statements hold for any tree T of order n(T)≥2 with s(T)=l(T).
(ⅰ) γtI(T)≥γ(T)+Δ(T).
(ⅱ) γtI(T)=γ(T)+Δ(T) if and only if T∈T.
Proof. By Theorem 3.2 and the fact that s(T)=l(T)≥Δ(T), it follows that γtI(T)≥γ(T)+s(T)=γ(T)+l(T)≥γ(T)+Δ(T). Hence, (ⅰ) follows. Now, we proceed to prove (ⅱ). First, we suppose that γtI(T)=γ(T)+Δ(T). From the previous inequality chain we obtain that l(T)=Δ(T) and that γtI(T)=γ(T)+s(T). The equality l(T)=Δ(T) implies that V(T)∖{v}⊆V≤2(T), where v∈V(T) is a vertex of maximum degree. Let g(W0,W1,W2) be a γtI(T)-function which satisfies Lemma 3.1 (notice that n(T)≥4 because s(T)=l(T)). Thus, S(T)∖{v}⊆V≤2(T)⊆W0∪W1. In addition, as γtI(T)=γ(T)+s(T), then we have equalities through the inequality chain given in the proof of Theorem 3.2. In particular, we have that |W2∪(W1∖L(T))|=γ(T), which implies that D=W2∪(W1∖L(T)) is a γ(T)-set. In order to deduce that T∈T, we first show that d(v,h)≤3 for every h∈L(T). For this purpose, we suppose that there exists a leaf h′ such that d(v,h′)≥4. Let us consider the path v=v0v1⋯vr=h′ (r≥4). Since V(T)∖{v}⊆V≤2(T), it follows by Lemma 3.1 that V(T)∖{v}⊆W0∪W1, and without loss of generality, we can assume that vr,vr−1,vr−3∈W1, vr−2∈W0 and vr−4∈W1∪W2. Hence, D∖{vr−3} is a dominating set of T of cardinality |D|−1=γ(T)−1, a contradiction. Therefore, d(v,h)≤3 for every h∈L(T), as required. Now, we suppose that T is not a subdivided star and that v∉S(T). Hence, there exists a path vv1v2v3 with v3∈L(T). As above, and without loss of generality, we can assume that v∈W1∪W2 and that v1∈W0. Since S(T)⊆W1 and N(v)∩W1≠∅, it follows that D∖{v} is a dominating set of T of cardinality |D|−1=γ(T)−1, a contradiction. Therefore, either T is a subdivided star or T can be obtained from a star K1,n by subdividing exactly n−1 edges at most twice (recall that s(T)=l(T)). That is, T∈T, as required. Finally, it is straightforward to observe that if T∈T, then γtI(T)=γ(T)+Δ(T).
Theorem 3.4. If T is a tree of order n(T)≥2, then
γ×2(T)≤n(T)−γ(T)+s(T). |
Proof. Let S be a 2-packing set of T of cardinality ρ(T) such that |S∩L(T)| is maximum. By the maximality of |S∩L(T)|, it follows that |N(x)∩L(T)∩S|=1 for every x∈S(T), which implies that |S∩L(T)|=|S(T)|. Let D=(V(T)∖S)∪L(T). Observe that |D|=n(T)−|S|+|S∩L(T)|=n(T)−ρ(T)+s(T). Now, we proceed to prove that D is a DDS of T. As V(T)∖D⊆S∖L(T), it is easy to observe that |N(x)∩D|≥2 for any vertex x∈V(T)∖D. Now, if x∈D, then N(x)∩D≠∅ because N(x)⊈S. Hence, D is a DDS of T, as desired. Therefore, γ×2(T)≤|D|=n(T)−ρ(T)+s(T). Finally, the result follows due to the fact that γ(T)=ρ(T) for any tree T (see [15]).
The following result is an immediate consequence of the previous theorem and the fact that γtI(T)≤γ×2(T) for any nontrivial tree T.
Corollary 3.5. If T is a tree of order n(T)≥2, then
γtI(T)≤n(T)−γ(T)+s(T). |
To see the tightness of the bounds given in Theorem 3.4 and Corollary 3.5 we consider for instance the path P3k+1 with k≥1. For this particular tree, it is easy to deduce that γ×2(P3k+1)=γtI(P3k+1)=2k+2=n(P3k+1)−γ(P3k+1)+s(P3k+1).
The main goal of this section is to show that for any tree T of order n(T)≥2,
n(T)+γ(T)+s(T)−l(T)+12≤γtI(T)≤n(T)+γ(T)+l(T)2. | (4.1) |
The previous upper bound is a direct consequence of the well-known relationship γtI(T)≤γ×2(T) and the inequality γ×2(T)≤(n(T)+γ(T)+l(T))/2 given by Cabrera-Martínez in [10]. In order to deduce the lower bound, we first need to state the following useful lemmas.
Lemma 4.1. [16] The following statements hold for any tree T of order n(T)≥2.
(ⅰ) If T is obtained from any nontrivial tree T′ by attaching a path P2 to any vertex u∈S(T′)∪SS(T′), then γ(T)=γ(T′)+1.
(ⅱ) If T is obtained from any nontrivial tree T′ by attaching a path P3 to any vertex u∈V(T′), then γ(T)=γ(T′)+1.
Lemma 4.2. If T is a tree obtained from any nontrivial tree T′ by attaching a path P3 to any vertex u∈V(T′), then
γtI(T)≥γtI(T′)+2. |
Proof. Let T be a tree obtained from T′ by adding the path ud−1udud+1 and the edge uud−1, where u∈V(T′). By Lemma 3.1-(b) and the fact that ud∈S(T), ud+1∈L(T) and ud−1∈V≤2(T), there exists a γtI(T)-function f such that f(ud)=f(ud+1)=1 and f(ud−1)≤1. We next define a function f′ on T′ as follows.
(a) If f(ud−1)=0, then f′(x)=f(x) whenever x∈V(T′).
(b) If f(ud−1)=1 and f(u)=0, then f′(u)=1 and f′(x)=f(x) whenever x∈V(T′)∖{u}.
(c) If f(ud−1)=1 and f(u)>0, then f′(u′)=1 for some u′∈N(u)∖{ud−1} and f′(x)=f(x) whenever x∈V(T′)∖{u′}.
It is straightforward that the function f′, defined through any of the three previous options, is a TIDF on T′ with weight ω(f)−2. Hence, γtI(T′)+2≤ω(f′)+2=ω(f)=γtI(T), as desired.
Theorem 4.3. If T is a tree of order n(T)≥2, then
γtI(T)≥n(T)+γ(T)+s(T)−l(T)+12. |
Proof. We proceed by applying induction on the order of T. If n(T)∈{2,3}, then it is straightforward that γtI(T)≥(n(T)+γ(T)+s(T)−l(T)+1)/2. These particular cases establish the base cases. Let T be a tree of order at least four, and P=u1u2⋯ud+1 be a diametrical path in T, where d=diam(T). Observe that u1,ud+1∈L(T). From now on, we consider that each tree T′ with n(T′)<n(T) satisfies that γtI(T′)≥(n(T′)+γ(T′)+s(T′)−l(T′)+1)/2. Now, we differentiate the following cases.
Case 1. degT(ud)≥3. In this case, we consider that T′=T−{ud+1}. Let g(W0,W1,W2) be a γtI(T)-function that satisfies the conditions of Lemma 3.1. By the condition (a) it follows that ud∈W2 and, without loss of generality, we can assume that ud+1∈W0. So, the function g restricted to V(T′) is a TIDF on T′. Hence, γtI(T′)≤g(V(T′))=ω(g)−g(ud+1)=γtI(T). In addition, as there exists a γ(T)-set containing no leaves, it is straightforward that γ(T′)=γ(T). Thus, by the inequalities above, the induction hypothesis and the fact that n(T)=n(T′)+1, s(T)=s(T′) and l(T)=l(T′)+1, we have the following desired result.
γtI(T)≥γtI(T′)≥n(T′)+γ(T′)+s(T′)−l(T′)+12=n(T)−1+γ(T)+s(T)−(l(T)−1)+12=n(T)+γ(T)+s(T)−l(T)+12. |
As can be seen from the proof above, the position of vertex ud∈S(T) is not relevant. Hence, we may henceforth assume that Ss(T)=∅.
Case 2. degT(ud)=2 and degT(ud−1)≥3. Let T′=T−{ud+1,ud}. Let g(W0,W1,W2) be a γtI(T)-function such that |W1∩L(T)| is maximum.
Subcase 2.1. The function g restricted to V(T′) is a TIDF on T′. As g(ud+1)+g(ud)=2, it follows that γtI(T′)≤g(V(T′))=ω(g)−(g(ud+1)+g(ud))=γtI(T)−2. Also, as degT(ud−1)≥3, it follows that ud−1∈S(T′)∪SS(T′), which leads to γ(T)=γ(T′)+1 by Lemma 4.1-(ⅰ). Thus, by the inequalities above, the induction hypothesis and the fact that n(T)=n(T′)+2, s(T)=s(T′)+1 and l(T)=l(T′)+1, we have the following desired result.
γtI(T)≥γtI(T′)+2≥n(T′)+γ(T′)+s(T′)−l(T′)+12+2=n(T)−2+γ(T)−1+s(T)−1−(l(T)−1)+12+2>n(T)+γ(T)+s(T)−l(T)+12. |
Subcase 2.2. The function g restricted to V(T′) is not a TIDF on T′. Let us observe that N(ud−1)∖{ud,ud−2}⊂S(T)∪L(T) and ud−1∉Ss(T). If degT(ud−1)≥4 or g(ud−1)=1, then g(N(ud−1)∖{ud,ud−2})>0, and as a consequence, we obtain that g restricted to V(T′) is a TIDF on T′, a contradiction. Therefore, degT(ud−1)=3 and g(ud−1)≠1.
First, we consider that g(ud−1)=0. In this case, it follows that ud−1∈SS(T), |V(Tud−1)|=5 and V(Tud−1)∖{ud−1}⊆W1 (recall that |W1∩L(T)| is maximum). Let T″=T−V(Tud−1). Notice that g restricted to V(T″) is a TIDF on T″, which implies that γtI(T″)≤g(V(T″))=ω(g)−g(V(Tud−1))=γtI(T)−4. Also, it is straightforward that γ(T)≤γ(T″)+2. Thus, by the inequalities above, the induction hypothesis and the fact that n(T)=n(T″)+5, s(T)≤s(T″)+2 and l(T)≥l(T″)+1, we have the following desired result.
γtI(T)≥γtI(T″)+4≥n(T″)+γ(T″)+s(T″)−l(T″)+12+4=n(T)−5+γ(T)−2+s(T)−2−(l(T)−1)+12+4=n(T)+γ(T)+s(T)−l(T)+12. |
Finally, we consider that g(ud−1)=2. This implies that ud−1∈S(T), |V(Tud−1)|=4 and ud,ud+1∈W1. If g(N[ud−2]∖{ud−1})>0, then the function g′(W′0,W′1,W′2), defined as g′(x)=1 whenever x∈V(Tud−1) and g′(x)=g(x) otherwise, is a γtI(T)-function satisfying that |W′1∩L(T)|>|W1∩L(T)|, which is a contradiction. Therefore, N[ud−2]∖{ud−1}⊆W0, which implies that degT(ud−2)=2 as a consequence of the maximality of |W1∩L(T)|. Let T∗=T−V(Tud−2). Since ud−2∈W0, it follows that g restricted to V(T∗) is a TIDF on T∗, which implies that γtI(T∗)≤g(V(T∗))=ω(g)−g(V(Tud−2))=γtI(T)−4. Notice that γ(T)≤γ(T∗)+2, n(T)=n(T∗)+5, s(T)≤s(T∗)+2 and l(T)≥l(T∗)+1. Hence, and proceeding analogously to the previous case (g(ud−1)=0), it follows the following desired inequality.
γtI(T)≥n(T)+γ(T)+s(T)−l(T)+12. |
Case 3. degT(ud−1)=degT(ud)=2. In this case, let T′=T−{ud−1,ud,ud+1}. By Lemmas 4.2 and 4.1-(ⅱ) we have that γtI(T)≥γtI(T′)+2 and γ(T)=γ(T′)+1, respectively. Now, we observe that s(T′)≥s(T)−1 and l(T′)≤l(T). Next, we analyse the following two subcases.
Subcase 3.1. s(T′)≥s(T) or l(T′)≤l(T)−1. In this subcase, by the previous inequalities, the induction hypothesis and the fact that n(T)=n(T′)+3, we have the following desired result.
γtI(T)≥γtI(T′)+2≥n(T′)+γ(T′)+s(T′)−l(T′)+12+2≥n(T)−3+γ(T)−1+s(T)−l(T)+12+2=n(T)+γ(T)+s(T)−l(T)+12. |
Subcase 3.2. s(T′)=s(T)−1 and l(T′)=l(T). These previous conditions lead to degT(ud−2)=2 and ud−3∈S(T). Let T″=T−{ud,ud+1} and let g(W0,W1,W2) be a γtI(T)-function such that |W1∩{ud−2,ud−1,ud,ud+1}| is maximum. Since ud−3∈W1∪W2, we can assume that ud−2,ud,ud+1∈W1 and ud−1∈W0. Notice that the function g′, defined by g′(ud−1)=1 and g′(x)=g(x) if x∈V(T″)∖{ud−1}, is a TIDF on T″. Therefore, γtI(T″)+1≤ω(g′)+1=ω(g)=γtI(T). In addition, we can deduce that γ(T″)=γ(T) because ud−3∈S(T)∩S(T″). By the previous inequalities, the induction hypothesis and the fact that n(T)=n(T″)+2, s(T″)=s(T) and l(T″)=l(T), we have the following desired result.
γtI(T)≥γtI(T″)+1≥n(T″)+γ(T″)+s(T″)−l(T″)+12+1=n(T)−2+γ(T)+s(T)−l(T)+12+1=n(T)+γ(T)+s(T)−l(T)+12. |
Therefore, and as consequence of the three cases above, the proof follows.
In order to show the sharpness of the lower bound given in Theorem 4.3, we need to define the next family of trees. Given an integer r≥2, the tree Gr is constructed from the path P2r+1=v1v2⋯v2r+1 and the path P1 by taking one copy of P2r+1 and r+1 copies of P1 and adding edges between the vertex v2i+1 and the i-th copy of P1 with i∈{0,…,r}. In Figure 1 we can show the tree G2. Observe that n(Gr)=3r+2, γ(Gr)=s(Gr)=l(Gr)=r+1 and γtI(Gr)=2r+2. Therefore,
γtI(Gr)=2r+2=(3r+2)+(r+1)+12=n(Gr)+γ(Gr)+s(Gr)−l(Gr)+12. |
In this article we studied the total Italian domination number of nontrivial trees. In particular, we obtained new lower and upper bounds on this domination parameter in terms of order, domination number, number of support vertices and number of leaves of a nontrivial tree. In addition, we discussed some extreme cases.
The problem of computing γtI(T) for trees T can be solved in polynomial time according to [4]. Nevertheless, through the relationships given in this article, new results can be obtained, as well as defining some new open problems. The following theorem provides two consequences derived from inequality chain (3.1) and the bound γ(T)≥(n(T)−l(T)+2)/3 given in [17].
Theorem 5.1. The following statements hold for any tree T of order n(T)≥3.
(ⅰ) γ(T)≤n(T)+s(T)3.
(ⅱ) γtI(T)≤2n(T)+l(T)+3s(T)−23.
Proof. By Theorem 3.2 and Corollary 3.5, we have that 2γ(T)≤γtI(T)≤n(T)−γ(T)+s(T). From this previous inequality chain we deduce that γ(T)≤(n(T)+s(T))/3, which completes the proof of (ⅰ).
Finally, we proceed to prove (ⅱ). By Corollary 3.5 and the fact that γ(T)≥(n(T)−l(T)+2)/3 (see [17]), it follows that
γtI(T)≤n(T)−γ(T)+s(T)≤n(T)−n(T)−l(T)+23+s(T)=2n(T)+l(T)+3s(T)−23, |
which completes the proof.
In addition, the bound γtI(T)≥⌈(2(n(T)−l(T)+3))/3⌉ given in [4] can also be deduced as a consequence of the relationships γtI(T)≥2γ(T) and γ(T)≥(n(T)−l(T)+2)/3 and the characterization of the previous equality given by Lemańska [17].
Finally, we propose some open problems that arise from the results obtained in the article.
(ⅰ) Characterize the trees with γtI(T)=2γ(T).
(ⅱ) Characterize the trees with γtI(T)=n(T)−γ(T)+s(T) or γ×2(T)=n(T)−γ(T)+s(T).
(iii) Characterize the trees attaining the bounds given in the inequality chain (4.1).
(iv) Obtain new lower and upper bounds on the total Italian domination number of trees.
The authors would like to thank the anonymous reviewers for the careful reading of the manuscript, and for all the suggestions which contributed to improve the quality and presentation of the paper.
The authors declare no conflict of interest.
[1] | T. W. Haynes, S. T. Hedetniemi, P. J. Slater, Fundamentals of domination in graphs, Chapman and Hall/CRC Pure and Applied Mathematics Series, Marcel Dekker, Inc. New York, 1998. |
[2] | T. W. Haynes, S. T. Hedetniemi, P. J. Slater, Fundamentals of domination in graphs: Advanced topics, Chapman & Hall/CRC Pure and Applied Mathematics, Taylor & Francis, 1998. |
[3] |
S. C. García, A. Cabrera-Martínez, F. A. Hernández Mira, I. G. Yero, Total Roman {2}-domination in graphs, Quaest. Math., 44 (2022), 411–434. https://doi.org/10.2989/16073606.2019.1695230 doi: 10.2989/16073606.2019.1695230
![]() |
[4] |
H. A. Ahangar, M. Chellali, S. M. Sheikholeslami, J. C. Valenzuela-Tripodoro, Total Roman {2}-dominating functions in graphs, Discuss. Math. Graph Theory, 42 (2022), 937–958. https://doi.org/10.7151/dmgt.2316 doi: 10.7151/dmgt.2316
![]() |
[5] |
H. Abdollahzadeh Ahangar, M. Chellali, M. Hajjari, S. M. Sheikholeslami, Further progress on the total Roman {2}-domination number of graphs, Bull. Iranian Math. Soc., 48 (2022), 1111–1119. https://doi.org/10.1007/s41980-021-00565-z doi: 10.1007/s41980-021-00565-z
![]() |
[6] |
A. Cabrera-Martínez, S. C. García, J. A. Rodríguez-Velázquez, Double domination in lexicographic product graphs, Discrete Appl. Math., 284 (2020), 290–300. https://doi.org/10.1016/j.dam.2020.03.045 doi: 10.1016/j.dam.2020.03.045
![]() |
[7] |
P. Chakradhar, P. V. S. Reddy, Algorithmic aspects of total Roman {2}-domination in graphs, Commun. Comb. Optim., 7 (2022), 183–192. https://doi.org/10.22049/CCO.2021.27063.1189 doi: 10.22049/CCO.2021.27063.1189
![]() |
[8] |
S. M. Sheikholeslami, L. Volkmann, Nordhaus-Gaddum type inequalities on the total Italian domination number in graphs, RAIRO Oper. Res., 56 (2022), 2235–2243. https://doi.org/10.1051/ro/2022108 doi: 10.1051/ro/2022108
![]() |
[9] | F. Harary, T. W. Haynes, Double domination in graphs, Ars Combin., 55 (2000), 201–213. |
[10] |
A. Cabrera-Martínez, New bounds on the double domination number of trees, Discrete Appl. Math., 315 (2022), 97–103. https://doi.org/10.1016/j.dam.2022.03.022 doi: 10.1016/j.dam.2022.03.022
![]() |
[11] |
A. Cabrera-Martínez, Some new results on the k-tuple domination number of graphs, RAIRO Oper. Res., 56 (2022), 3491–3497. https://doi.org/10.1051/ro/2022159 doi: 10.1051/ro/2022159
![]() |
[12] |
A. Cabrera-Martínez, J. A. Rodríguez-Velázquez, A note on double domination in graphs, Discrete Appl. Math., 300 (2021), 107–111. https://doi.org/10.1016/j.dam.2021.05.011 doi: 10.1016/j.dam.2021.05.011
![]() |
[13] |
M. Hajian, N. J. Rad, A new lower bound on the double domination number of a graph, Discrete Appl. Math., 254 (2019), 280–282. https://doi.org/10.1016/j.dam.2018.06.009 doi: 10.1016/j.dam.2018.06.009
![]() |
[14] |
N. A. A. Aziz, N. J. Rad, H. Kamarulhaili, A note on the double domination number in maximal outerplanar and planar graphs, RAIRO Oper. Res., 56 (2022), 3367–3371. https://doi.org/10.1051/ro/2022150 doi: 10.1051/ro/2022150
![]() |
[15] | A. Meir, J. W. Moon, Relations between packing and covering numbers of a tree, Pacific J. Math., 61 (1975), 225–233. |
[16] |
A. Cabrera-Martínez, A. C. Peiró, On the {2}-domination number of graphs. AIMS Math., 7 (2022), 10731–10743. https://doi.org/10.3934/math.2022599 doi: 10.3934/math.2022599
![]() |
[17] |
M. Lemanska, Lower bound on the domination number of a tree, Discuss. Math. Graph Theory, 24 (2004), 165–169. https://doi.org/10.7151/dmgt.1222 doi: 10.7151/dmgt.1222
![]() |
1. | Abel Cabrera-Martínez, An improved upper bound on the domination number of a tree, 2024, 343, 0166218X, 44, 10.1016/j.dam.2023.10.013 |