1. Introduction
Let
be a finite simple graph of order
and size
. For a set
, we define
and
. As usual,
denotes the graph obtained from
G by removing all the vertices in
D and all the edge incidents with a vertex in
D. Analogously, the graph obtained from
G by removing all the edges in
will be denoted by
. Given a vertex
v of
G,
and
represent the
open neighborhood and the
closed neighborhood of vertex
v; that is,
and
. The
degree of a vertex
v, denoted by
, is the cardinality of
. A vertex of degree one is called a
leaf, and its neighbor is called a
support vertex. Let
and
, where
and
. A
strong leaf is a leaf at distance two from another leaf. The set of strong leaves is denoted by
. A connected graph
G is a cactus graph if each edge of
G is contained in at most one cycle. We will use the notation
,
and
for the path graphs, cycle graphs and star graphs of order
n, respectively. For any other graph theory terminology, we follow the book [
1].
Domination in graphs is one of the most popular and highly investigated topics in the area of graph theory. A set
is called a
dominating set of
G if
for every
. The
domination number of
G, denoted by
, is defined as
. A compendium of the main results obtained on domination theory in graphs until 1998 can be found in books [
1,
2].
In the last decades, one interesting research activity on domination in graphs has been the study of the parameters associated with different variants of dominating sets in graphs. These domination parameters will depend on conditions that can be imposed on the dominating set D, on the set or on the “method” by which vertices in are dominated.
In this article, we study three domination parameters in cactus graphs, which present a certain symmetry due to the conditions that are imposed on each of them. In particular, and as we will show below, these parameters are related to dominating sets whose complements are independent sets, which are nowadays very common research topics in the graph theory community. Next, we define our three domination parameters of interest:
The total outer-independent domination number of
G, denoted by
, is the minimum cardinality among all
total outer-independent dominating sets (TOIDSs) of
G. In this case, a set
is a TOIDS of
G if
for every vertex
, and
is an independent set of
G. A
-
set is a TOIDS of
G of cardinality
. This domination parameter was introduced by Soner et al. [
3]. For recent results on the total outer-independent domination in graphs, we cite [
4,
5,
6].
The double outer-independent domination number of
G, denoted by
, is the minimum cardinality among all
double outer-independent dominating sets (DOIDSs) of
G. In this case, a set
is a DOIDS of
G if
for every vertex
, and
is an independent set of
G. A
-
set is a DOIDS of
G of cardinality
. The study of this domination parameter was initiated in [
7]. Some recent and excellent results on this concept can be found, for example, in [
6,
8,
9].
The 2-outer-independent domination number of
G, denoted by
, is the minimum cardinality among all
2-outer-independent dominating sets (2OIDSs) of
G. In this case, a set
is a 2OIDS of
G if
for every vertex
, and
is an independent set of
G. A
-
set is a 2OIDS of
G of cardinality
. This parameter was introduced in [
10] and studied further in [
11].
As a consequence of the definitions above, we deduce that if
D is a
-set or a
-set, then
. In the same way, if
W is a
-set or a
-set, then
. These facts will be very useful tools throughout the article.
Figure 1 shows a cactus graph
G with
,
and
.
The following theorem, due to Krzywkowski, establishes lower and upper bounds on the previous three outer-independent domination-related parameters for trees.
Theorem 1. The following bounds hold for any tree T of order .
- (i)
(obtained in [
11]
). - (ii)
(obtained in [
12]
and [
13]
, respectively). - (iii)
(obtained in [
14]
and [
8]
, respectively).
In this paper, we first extend the upper bounds given in the previous theorem for the case of the cactus graphs. We prove that for any cactus graph G of order at least four with cycles,
;
;
.
Moreover, we briefly address the particular case of trees. In particular, we characterize the nontrivial trees T with and we show that for any tree T with . This previous lower bound covers an existing gap for the 2-outer-independent domination number of a tree, in relation to the other two parameters of interest (see the lower bounds given in Theorem 1 for and ). Finally, we give a constructive characterization of the trees that satisfy the equality above.
2. New Upper Bounds
We begin this section by extending the upper bound given in Theorem 1-(i) for the case of the cactus graphs. Before, let us recall that for any cycle of order .
Theorem 2. If G is a nontrivial cactus graph with cycles, then Proof. Let G be a nontrivial cactus graph. The proof is by induction on the size . If , then it is straightforward that the result follows. These establish the base cases. Let us assume that and that for each nontrivial cactus graph with . If G is a tree of order at least four, then the result follows by Theorem 1-(i). On the other hand, if G is a cycle of order , then the result follows by the fact that . Henceforth, we will assume that G is a cactus graph other than a cycle or a tree. Hence, G contains at least one cycle as a proper subgraph. Let C be any cycle of G. Also, let such that and . Let . Observe that is a cactus graph with , and . By the induction hypothesis, we have the inequality . In addition, we observe that or . Now, we can distinguish two cases as follows:
The bound given in the theorem above is tight. For instance, it is achieved for the nontrivial trees attaining the upper bound given in Theorem 1-(i) (see [
11]).
In the following result, we extend the upper bound given in Theorem 1-(ii) for the case of the cactus graphs. Before, let us recall that for any cycle of order .
Theorem 3. If G is a cactus graph of order at least four with cycles, then Proof. Let G be a cactus graph of order at least four. We proceed by induction on the size . If , then G is either the path or the star . In both cases, the inequality holds by using the upper bound given in Theorem 1-(ii). This establishes the base case. Let us assume that and that for each cactus graph of order at least four such that .
If G is a tree or a cycle, then the result follows by Theorem 1-(ii) or by the fact that , respectively. Henceforth, we will assume that G is a cactus graph other than a cycle or a tree. Hence, G contains at least one cycle as a proper subgraph. Let C be any cycle of G. Also, let such that and . Let . Observe that is a cactus graph with , and . By the induction hypothesis, we have the inequality . Now, we proceed to show that . For this purpose, we consider a -set . Observe that is a TOIDS of G, which implies that , as desired. In addition, we observe that or . Now, we can distinguish two cases as follows:
Case 1:
. It is easy to check that
. Therefore, by the previous inequalities and the induction hypothesis, we obtain the following desired result:
Case 2:
. In this case, it follows that
. Thus, by the inequality above and the induction hypothesis, we obtain the following desired result:
From the two cases above, the proof follows. □
The bound given in the theorem above is tight. For instance, it is achieved for the nontrivial trees attaining the upper bound given in Theorem 1-(ii) (see [
13]).
Next, we extend the upper bound given in Theorem 1-(iii) for the case of the cactus graphs.
Theorem 4. If G is a cactus graph of order at least three with cycles, then Proof. Let G be a cactus graph with . We proceed by induction on the size . If , then it is easy to check that the result follows. These establish the base cases. We assume that and that for each cactus graph of order such that .
If G is a tree or a cycle, then the result follows by Theorem 1-(iii) or by the fact that , respectively. Henceforth, we will assume that G is a cactus graph other than a cycle or a tree. Hence, G contains at least one cycle as a proper subgraph. Let C be any cycle of G. Also, let such that and . Let . Observe that is a cactus graph with , and . By the induction hypothesis, we have the inequality . In addition, we observe that or . Now, we can distinguish two cases as follows:
Case 1:
. It is easy to check that
and that
. Let
be a
-set. Since
, it follows that
. As a consequence, we deduce that
is also a DOIDS of
G. Hence,
. Therefore, by the induction hypothesis and the previous inequalities, we obtain the following desired result:
Case 2:
. Observe that
. Now, let
be a
-set. Observe that
is a DOIDS of
G, which implies that
. Hence, by the inequalities above and the induction hypothesis, we obtain the following desired result:
From the two cases above, the proof follows. □
The bound given in the theorem above is tight. For instance, it is achieved for the nontrivial trees attaining the upper bound given in Theorem 1-(iii) (see [
8]).
The Particular Case of Trees
In this subsection, we first address that gap for the 2-outer-independent domination number of a tree of order at least three.
Theorem 5. For any tree T of order , Proof. Let T be a tree of order at least three. We proceed by induction on the order . If , then the result follows. These establish the base cases. We assume that and that for each tree with . Let be a diametral path in T, and we consider the following three cases:
Let be the family of trees T that can be obtained from a sequence of trees , with and . If , then for each subscript , the tree can be obtained from by one of the next two operations:
Operation : Add a path and join it to a vertex .
Operation : Add a path and join one of its leaves to a vertex which is in some -set.
Now, we proceed to prove that every tree in achieves equality in the lower bound given in Theorem 5.
Lemma 1. Let T be a tree of order . If , then .
Proof. Let T be a tree of order at least three. We proceed by induction on the number of operations required to construct the tree T. If , then and , as required. This establishes the base case. We now assume that is an integer and that each tree with satisfies that . Since , it follows that T can be obtained from a tree with by one of the Operations or . Next, we consider the next two cases.
Case 1: T is obtained from by Operation . In this case, T is obtained from by adding a vertex and the edge , where . Observe that for any -set , the set is a 2OIDS of T. This implies that . By using the lower bound given in Theorem 5, the induction hypothesis and the fact that , and , it follows that . Hence, as desired.
Case 2: T is obtained from by Operation . In this case, T is obtained from by adding the path and the edge , where for some -set . Observe that is a 2OIDS of T. This implies that . By using the lower bound given in Theorem 5, the induction hypothesis and the fact that and , we have that . Hence, we have that as desired. □
We next show that every tree T of order at least three satisfying equality belongs to the family .
Lemma 2. Let T be a tree of order . If , then .
Proof. We proceed by induction on the order of a tree T, which satisfies that . If then , which belongs to . We assume that and that every tree with and satisfies that . Now, we prove that . For this, we root the tree T at a leaf belonging to a diametral path . We consider the following cases:
Case 1:
. Let us consider the subtree
. Let
D be a
-set. As
and
is an independent set of
T, we have that
,
and
. By the previous conditions, it is easy to deduce that
is a 2OIDS of
, which implies that
. By using the lower bound given in Theorem 5 for the tree
, the hypothesis
and the fact that
,
and
, it follows that
Hence, we have that
, and by the induction hypothesis, it follows that
. Since
, we have that
T can be obtained from
by Operation
, which implies that
, as desired.
Case 2:
. Let us consider the subtree
. Let
D be a
-set such that
and
(such a set
D exists because
,
is an independent set of
T and
). Observe that
is a 2OIDS of
. This implies that
. If
, then
and
, which implies that
, which contradicts the bound given in Theorem 5. Hence,
, and as a consequence, it follows that
and
. Therefore, by using the lower bound given in Theorem 5 for the tree
, the hypothesis
and the previous equalities, we have that
Hence, we have that
, and by the induction hypothesis, it follows that
. Moreover, it follows that
, which implies that
is a
-set containing vertex
. Since
, we have that
T can be obtained from tree
by Operation
, which implies that
, as desired. □
As an immediate consequence of Lemmas 1 and 2, we have the following characterization:
Theorem 6. Let T be a tree of order at least three. Then, if and only if .
Let be a tree obtained by subdividing the central edge of a double star exactly once. It is easy to see that the tree can only be obtained from the path by applying Operation . On the other hand, observe that the path can only be obtained from path by applying Operation . Therefore, Operations and are required in the characterization above.
Finally, we show an interesting result, which is a consequence of Theorem 1. Observe that, by definition, it is easy to deduce that if G is a graph with , then . However, characterizing the graphs G with that satisfy the above equality remains a problem to be solved. Next, we give a solution to this previous problem considering that G is a nontrivial tree.
Theorem 7. Let T be a nontrivial tree. Then, if and only if .
Proof. If
, then the equality
holds. From now on, we suppose that
. We only need to prove that
(recall that
). If
, then we are done. Let us consider that
. From Theorem 1, and considering the upper bound given in (ii) and the lower bound given in (iii), we obtain the following inequality chain:
In particular, it follows that
, as desired. Therefore, the proof is complete. □