1. Introduction
A generator of a metric space is a set S of points in the space with the property that every point of the space is uniquely determined by its distances from the elements of S. Given a simple and connected graph , we consider the function , where is the length of the shortest path between u and v and is the set of non-negative integers. Clearly, is a metric space, i.e., satisfies for all , for all and for all . A vertex is said to distinguish two vertices x and y if . A set is said to be a metric generator for G if any pair of vertices of G is distinguished by some element of S.
Metric generators were introduced by Blumental [
1] in the general context of metric spaces. They were later introduced in the context of graphs by Slater in [
2], where metric generators were called locating sets, and, independently, by Harary and Melter in [
3], where metric generators were called resolving sets. Applications of the metric dimension to the navigation of robots in networks are discussed in [
4] and applications to chemistry in [
5,
6]. This invariant was studied further in a number of other papers including, for instance [
7,
8,
9,
10,
11,
12,
13,
14,
15,
16,
17,
18,
19,
20].
As pointed out by Okamoto et al. in [
21], there exist applications where only neighboring vertices need to be distinguished. Such applications were the basis for the introduction of the local metric dimension. A set
is said to be a local metric generator for
G if any pair of adjacent vertices of
G is distinguished by some element of
S. A minimum local metric generator is called a local metric basis and its cardinality the local metric dimension of
G, denoted by
. Additionally, Jannesari and Omoomi [
16] introduced the concept of adjacency resolving sets as a result of considering the two-distance in
, which is defined as
for any two vertices
. A set of vertices
such that any pair of vertices of
is distinguished by an element
s in
considering the two-distance in
is called an adjacency generator for
G. If we only ask
to distinguish the pairs of adjacent vertices, we call
a local adjacency generator. A minimum local adjacency generator is called a local adjacency basis, and the cardinality of any such basis is the local adjacency dimension of
G, denoted
.
The notion of simultaneous metric dimension was introduced in the framework of the navigation problem proposed in [
4], where navigation was studied in a graph-structured framework in which the navigating agent (which was assumed to be a point robot) moves from node to node of a “graph space”. The robot can locate itself by the presence of distinctively-labeled “landmark” nodes in the graph space. On a graph, there is neither the concept of direction, nor that of visibility. Instead, it was assumed in [
4] that a robot navigating on a graph can sense the distances to a set of landmarks. Evidently, if the robot knows its distances to a sufficiently large set of landmarks, its position on the graph is uniquely determined. This suggests the following problem: given a graph
G, what are the fewest number of landmarks needed and where should they be located, so that the distances to the landmarks uniquely determine the robot’s position on
G? Indeed, the problem consists of determining the metric dimension and a metric basis of
G. Now, consider the following extension of this problem, introduced by Ramírez-Cruz, Oellermann and Rodríguez-Velázquez in [
22]. Suppose that the topology of the navigation network may change within a range of possible graphs, say
. This scenario may reflect several situations, for instance the simultaneous use of technologically-differentiated redundant sets of landmarks, the use of a dynamic network whose links change over time, etc. In this case, the above-mentioned problem becomes determining the minimum cardinality of a set
S, which must be simultaneously a metric generator for each graph
,
. Therefore, if
S is a solution for this problem, then each robot can be uniquely determined by the distance to the elements of
S, regardless of the graph
that models the network at each moment. Such sets we called simultaneous metric generators in [
22], where, by analogy, a simultaneous metric basis was defined as a simultaneous metric generator of minimum cardinality, and this cardinality was called the simultaneous metric dimension of the graph family
, denoted by
.
In this paper, we recover Okamoto et al.’s observation that in some applications, it is only necessary to distinguish neighboring vertices. In particular, we consider the problem of distinguishing neighboring vertices in a multiple topology scenario, so we deal with the problem of finding the minimum cardinality of a set S, which must simultaneously be a local metric generator for each graph , .
Given a family
of connected graphs
on a common vertex set
V, we define a simultaneous local metric generator for
as a set
such that
S is simultaneously a local metric generator for each
. We say that a minimum simultaneous local metric generator for
is a simultaneous local metric basis of
and its cardinality the simultaneous local metric dimension of
, denoted by
or explicitly by
. An example is shown in
Figure 1, where the set
is a simultaneous local metric basis of
.
It will also be useful to define the simultaneous local adjacency dimension of a family of connected graphs on a common vertex set V, as the cardinality of a minimum set such that S is simultaneously a local adjacency generator for each . We denote this parameter as .
In what follows, we will use the notation , , , and for complete graphs, complete bipartite graphs, cycle graphs, empty graphs and path graphs of order n, respectively. Given a graph and a vertex the set is the open neighborhood of v, and the set is the closed neighborhood of v. Two vertices are true twins in G if , and they are false twins if . In general, two vertices are said to be twins if they are true twins or they are false twins. As usual, a set is a vertex cover for G if for every , or . The vertex cover number of G, denoted by , is the minimum cardinality of a vertex cover of G. The remaining definitions will be given the first time that the concept appears in the text.
The rest of the article is organized as follows. In
Section 2, we obtain some general results on the simultaneous local metric dimension of graph families.
Section 3 is devoted to the case of graph families obtained by small changes on a graph, while in
Section 4 and
Section 5, we study the particular cases of families of corona graphs and families of lexicographic product graphs, respectively. Finally, in
Section 6, we show that the problem of computing the simultaneous local metric dimension of graph families is NP-hard, even when restricted to families of graphs that individually have a (small) fixed local metric dimension.
2. Basic Results
Remark 1. Let be a family of connected graphs defined on a common vertex set V, and let . The following results hold:
- 1.
.
- 2.
.
- 3.
.
Proof. (1) is deduced directly from the definition of simultaneous local metric dimension. Let B be a simultaneous metric basis of , and let be two vertices not in B, such that in some . Since in there exists such that , B is a simultaneous local metric generator for , so (2) holds. Finally, (3) is obtained from the following facts: (a) the union of local metric generators for all graphs in is a simultaneous local metric generator for , which implies that ; (b) any vertex cover of is a local metric generator of , for every , which implies that . ☐
The inequalities above are tight. For example, the graph family
shown in
Figure 1 satisfies
, whereas
. Moreover, the family
shown in
Figure 2 satisfies
, whereas the family
shown in
Figure 3 satisfies
.
We now analyze the extreme cases of the bounds given in Remark 1.
Corollary 1. Let be a family of connected graphs on a common vertex set. If , then: As shown in
Figure 2, the converse of Corollary 1 does not hold. In general, the cases for which the upper bound
is reached are summarized in the next result.
Theorem 1. Let be a family of connected graphs on a common vertex set V. Then, if and only if for every , there exists a graph such that u and v are true twins in .
Proof. We first note that for any connected graph and any vertex , it holds that is a local metric generator for G. Therefore, if , then for any , the set is a simultaneous local metric basis of , and as a consequence, for every , there exists a graph , such that the set is not a local metric generator for , i.e., u and v are adjacent in and for every . Therefore, u and v are true twins in .
Conversely, if for every there exists a graph such that u and v are true twins in , then for any simultaneous local metric basis B of , it holds that or . Hence, all but one element of V must belong to B. Therefore, , which implies that . ☐
Notice that Corollary 1 is obtained directly from the previous result. Now, the two following results concern the limit cases of Item (1) of Remark 1.
Theorem 2. A family of connected graphs on a common vertex set V satisfies if and only if every graph in is bipartite.
Proof. If every graph in the family is bipartite, then for any , the set is a local metric basis of every , so .
Let us now consider a family
of connected graphs on a common vertex set
V such that
and assume that some
is not bipartite. It is shown in [
21] that
, so Item (1) of Remark 1 leads to
, which is a contradiction. Thus, every
is bipartite. ☐
Paths, trees and even-order cycles are bipartite. The following result covers the case of families composed of odd-order cycles.
Theorem 3. Every family composed of cycle graphs on a common odd-sized vertex set V satisfies , and any pair of vertices of V is a simultaneous local metric basis of .
Proof. For any cycle , the set , , is not a local metric generator, as the adjacent vertices and (subscripts taken modulo ) are not distinguished by v, so Item (1) of Remark 1 leads to . Moreover, any set is a local metric generator for every , as the single pair of adjacent vertices not distinguished by v is distinguished by , so that . ☐
The following result allows us to study the simultaneous local metric dimension of a family from the family of graphs composed by all non-bipartite graphs belonging to .
Theorem 4. Let be a family of graphs on a common vertex set V, not all of them bipartite. If is the subfamily of composed of all non-bipartite graphs belonging to , then: Proof. Since is a non-empty subfamily of , we conclude that Since any vertex of a bipartite graph G is a local metric generator for G, if is a simultaneous local metric basis of , then B is a simultaneous local metric generator for and, as a result, ☐
Some interesting situations may be observed regarding the simultaneous local metric dimension of some graph families versus its standard counterpart. In particular, the fact that false twin vertices need not be distinguished in the local variant leads to some cases where both parameters differ greatly. For instance, consider any family
composed of three or more star graphs having different centers. It was shown in [
22] that any such family satisfies
, yet by Theorem 2, we have that
.
Given a family of graphs on a common vertex set V, we define a simultaneous vertex cover of as a set , such that S is simultaneously a vertex cover of each . The minimum cardinality among all simultaneous vertex covers of is the simultaneous vertex cover number of , denoted by .
Theorem 5. For any family of connected graphs with common vertex set V,Furthermore, if for every there exists such that u and v are true twins in , then . Proof. Let be a simultaneous vertex cover of . Since is a simultaneous independent set of , we conclude that .
We now assume that for every , there exists , such that u and v are true twins in , and suppose, for the purpose of contradiction, that . In such a case, there exists a simultaneous local metric basis , which is not a simultaneous vertex cover of . Hence, there exist and such that , ergo . As a consequence, and are true twins in some graph , which contradicts the fact that C is a simultaneous local metric basis of . Therefore, the strict inequality does not hold, hence . ☐
3. Families Obtained by Small Changes on a Graph
Consider a graph G whose local metric dimension is known. In this section, we address two related questions:
If a series of small changes is repeatedly performed on , thus producing a family of consecutive versions of G, what is the behavior of with respect to ?
If several small changes are performed on in parallel, thus producing a family of alternative versions of G, what is the behavior of with respect to ?
Addressing this issue in the general case is hard, so we will analyze a number of particular cases. First, we will specify three operators that describe some types of changes that may be performed on a graph G:
Edge addition: We say that a graph is obtained from a graph G by an edge addition if there is an edge such that . We will use the notation .
Edge removal: We say that a graph is obtained from a graph G by an edge removal if there is an edge such that . We will use the notation .
Edge exchange: We say that a graph is obtained from a graph G by an edge exchange if there is an edge and an edge such that . We will use the notation .
Now, consider a graph G and an ordered k-tuple of operations , where . We define the class containing all graph families of the form , composed by connected graphs on the common vertex set , where for every . Likewise, we define the class containing all graph families of the form , composed by connected graphs on the common vertex set , where for every . In particular, if (, ) for every , we will write (, ) and (, ).
We have that performing an edge exchange on any tree T (path graphs included) either produces another tree or a disconnected graph. Thus, the following result is a direct consequence of this fact and Theorem 2.
Remark 2. For any tree T, any and any graph family , Our next result covers a large class of families composed by unicyclic graphs that can be obtained by adding edges, in parallel, to a path graph.
Remark 3. For any path graph , , any and any graph family , Proof. Every graph is either a cycle or a unicyclic graph. If the cycle subgraphs of every graph in the family have even order, then by Theorem 2. If contains at least one non-bipartite graph, then . We now proceed to show that in this case, . To this end, we denote by the vertex set of , where for every . We claim that is a simultaneous local metric generator for the subfamily composed by all non-bipartite graphs of . In order to prove this claim, consider an arbitrary graph , and let , be the edge added to to obtain G. We differentiate the following cases:
. In this case, G is an odd-order cycle graph, so is a local metric generator.
. In this case, G is a unicyclic graph where has degree three, has degree one and the remaining vertices have degree two. Consider two adjacent vertices . If u or v belong to the path from to , then distinguishes them. If both, u and v, belong to the cycle subgraph of G, then and . Thus, if distinguishes u and v, so does , otherwise does.
. This case is analogous to Case 2.
. In this case, G is a unicyclic graph where and have degree three, and have degree one and the remaining vertices have degree two. Consider two adjacent vertices . If u or v belong to the path from to (or to the path from to ), then (or ) distinguishes them. If both u and v belong to the cycle, then , , and . Thus, if distinguishes u and v, so does , otherwise distinguishes them, which means that also does.
According to the four cases above, we conclude that is a local metric generator for G, so it is a simultaneous local metric generator for . Thus, by Theorem 4, . ☐
Remark 4. Let , , be a cycle graph, and let e be an edge of its complement. If n is odd, thenOtherwise, Proof. Consider . We have that is bipartite for n even. If, additionally, is odd, then the graph is also bipartite, so . For every other case, . From now on, we assume that and proceed to show that . Note that is a bicyclic graph where and are vertices of degree three and the remaining vertices have degree two. We denote by and the two graphs obtained as induced subgraphs of , which are isomorphic to a cycle of order and a cycle of order , respectively. Since , we have that or . We assume, without loss of generality, that . Let are two vertices such that:
if is even, and ,
if is odd, , where is the only vertex such that .
We claim that is a local metric generator for . Consider two adjacent vertices . We differentiate the following cases, where the distances are taken in :
. It is simple to verify that is a local metric generator for , hence or .
and . In this case, and or .
. In this case, if , then , so and, consequently, .
According to the three cases above, is a local metric generator for , and as a result, the proof is complete. ☐
The next result is a direct consequence of Remarks 1 and 4.
Remark 5. Let , , be a cycle graph. If are two different edges of the complement of , then: 4. Families of Corona Product Graphs
Let
G and
H be two graphs of order
n and
, respectively. The corona product
is defined as the graph obtained from
G and
H by taking one copy of
G and
n copies of
H and joining by an edge each vertex from the
i-th copy of
H with the
i-th vertex of
G. Notice that the corona graph
is isomorphic to the join graph
. Given a graph family
on a common vertex set and a graph
H, we define the graph family:
Several results presented in [
23,
24] describe the behavior of the local metric dimension on corona product graphs. We now analyze how this behavior extends to the simultaneous local metric dimension of families composed by corona product graphs.
Theorem 6. In references [
23,
25],
Let G be a connected graph of order . For any non-empty graph H, As we can expect, if we review the proof of the result above, we check that if A is a local metric basis of , then A does not contain elements in . Therefore, any local metric basis of is a simultaneous local metric basis of . This fact and the result above allow us to state the following theorem.
Theorem 7. Let be a family of connected non-trivial graphs on a common vertex set V. For any non-empty graph H, Given a graph family
on a common vertex set and a graph family
on a common vertex set, we define the graph family:
The following result generalizes Theorem 7. In what follows, we will use the notation for the graph where and .
Theorem 8. For any family of connected non-trivial graphs on a common vertex set V and any family of non-empty graphs on a common vertex set, Proof. Let , and let be the vertex set of the graphs in , the copy of corresponding to , the i-th copy of and the i-th copy of .
We first need to prove that any satisfies . For any , let be a simultaneous local adjacency basis of . In order to show that is a simultaneous local metric generator for , we will show that X is a local metric generator for , for any and . To this end, we differentiate the following four cases for two adjacent vertices .
. Since
is an adjacency generator of
, there exists a vertex
such that
. Hence,
and
. If
, then for
,
, we have:
Now, if
,
, then we also take
, and we proceed as above.
and
. For
, we find that:
and
,
. In this case, for
, we have:
Hence,
X is a local metric generator for
, and since
and
are arbitrary graphs,
X is a simultaneous local metric generator for
, which implies that:
It remains to prove that
To do this, let
W be a simultaneous local metric basis of
, and for any
, let
. Let us show that
is a simultaneous adjacency generator for
. To do this, consider two different vertices
, which are adjacent in
, for some
. Since no vertex
distinguishes the pair
, there exists some
, such that
. Now, since
and
, we conclude that
, and consequently,
must be an adjacency generator for
; and since
is arbitrary,
is a simultaneous local adjacency generator for
. Hence, for any
,
. Therefore,
This completes the proof. ☐
The following result is a direct consequence of Theorem 8.
Corollary 2. For any family of connected non-trivial graphs on a common vertex set V and any family of non-empty graphs on a common vertex set,Furthermore, if every graph in has diameter two, then: Now, we give another result, which is a direct consequence of Theorem 8 and shows the general bounds of .
Corollary 3. For any family of connected graphs on a common vertex set V, and any family of non-empty graphs on a common vertex set , We now consider the case in which the graph H is empty.
Theorem 9. In reference [
24],
Let G be a connected non-trivial graph. For any empty graph H, The result above may be extended to the simultaneous scenario.
Theorem 10. Let be a family of connected non-trivial graphs on a common vertex set. For any empty graph H, Proof. Let B be a simultaneous local metric basis of . Since H is empty, any local metric generator of is a local metric generator for , so B is a simultaneous local metric generator for . As a consequence, .
Suppose that A is a simultaneous local metric basis of and . If there exists for the j-th copy of H in any graph , then the pairs of vertices of that are distinguished by x can also be distinguished by . As a consequence, the set obtained from A by replacing by each vertex is a simultaneous local metric generator for such that , which is a contradiction, so . ☐
Theorem 11. In reference [
24],
Let H be a non-empty graph. The following assertions hold. - 1.
If the vertex of does not belong to any local metric basis of , then for any connected graph G of order n, - 2.
If the vertex of belongs to a local metric basis of , then for any connected graph G of order ,
As for the previous case, the result above is extensible to the simultaneous setting.
Theorem 12. Let be a family of connected non-trivial graphs on a common vertex set V, and let be a family of non-empty graphs on a common vertex set. The following assertions hold.
- 1.
If the vertex of does not belong to any simultaneous local metric basis of , then: - 2.
If the vertex of belongs to a simultaneous local metric basis of , then:
Proof. As above, let , and let be the vertex set of the graphs in , the copy of corresponding to , the i-th copy of and the i-th copy of .
We will apply a reasoning analogous to the one used for the proof of Theorem 11 in [
24]. If
, then
, so the result holds. Assume that
, Let
be a simultaneous local metric basis of
, and let
. Note that
because
is the family of non-empty graphs and
does not distinguish any pair of adjacent vertices belonging to
. In order to show that
is a simultaneous local metric generator for
, we differentiate the following cases for two vertices
, which are adjacent in an arbitrary graph
:
. Since does not distinguish , there exists such that .
and . For , , we have .
and . For , we have .
Hence, X is a local metric generator for , and since and are arbitrary graphs, X is a simultaneous local metric generator for .
Now, we shall prove (1). If the vertex of
does not belong to any simultaneous local metric basis of
, then
for every
, and as a consequence,
Now, we need to prove that
In order to do this, let
W be a simultaneous local metric basis of
, and let
. Consider two adjacent vertices
in
. Since no vertex
distinguishes the pair
, there exists
such that
. Therefore, we conclude that
is a simultaneous local metric generator for
. Now, since
does not belong to any simultaneous local metric basis of
, we have that
and, as a consequence,
. Therefore,
and the proof of (1) is complete.
Finally, we shall prove (2). If the vertex of
belongs to a simultaneous local metric basis of
, then we assume that
for every
. Suppose that there exists
B such that
B is a simultaneous local metric basis of
and
. In such a case, there exists
such that the set
satisfies
. Now, since no vertex of
distinguishes the pairs of adjacent vertices belonging to
, the set
must be a simultaneous local metric generator for
. Therefore,
, which is a contradiction. Hence,
X is a simultaneous local metric basis of
, and as a consequence,
The proof of (2) is now complete. ☐
Corollary 4. Let G be a connected graph of order , and let , , be a family composed by complete bipartite graphs on a common vertex set . Then, Proof. For every , the set is a simultaneous local metric basis of , so . ☐
Lemma 1. In reference [
24],
Let H be a graph of radius . If , then the vertex of does not belong to any local metric basis of . Note that an analogous result holds for the simultaneous scenario.
Lemma 2. Let be a graph family on a common vertex set V, such that for every . Then, the vertex of does not belong to any simultaneous local metric basis of .
Proof. Let
B be a simultaneous local metric basis of
. We suppose that the vertex
v of
belongs to
B. Note that
if and only if there exists
, such that
for some
. If
, proceeding in a manner analogous to that of the proof of Lemma 1 as given in [
24], we take
such that
and a shortest path
. In such a case, for every
, we will have that
, which is a contradiction. Hence,
v does not belong to any simultaneous local metric basis of
. ☐
As a direct consequence of item (1) of Theorem 12 and Lemma 2, we obtain the following result.
Proposition 1. For any family of connected graphs on a common vertex set V and any graph family on a common vertex set such that for every , 5. Families of Lexicographic Product Graphs
Let
be a family of connected graphs with common vertex set
. For each
, let
be a family of graphs with common vertex set
. For each
, choose
and consider the family
. Notice that the families
can be represented in the following scheme where the columns correspond to the families
.
For a graph
and the family
, we define the lexicographic product of
and
as the graph
such that
and
if and only if
or
and
. Let
. We are interested in the simultaneous local metric dimension of the family:
The relation between distances in a lexicographic product graph and those in its factors is presented in the following remark.
Remark 6. If and are vertices of , then: We point out that the remark above was stated in [
26,
27] for the case where
for all
. By Remark 6, we deduce that if
, then two adjacent vertices
are not distinguished by
. Therefore, we can state the following remark.
Remark 7. If B is a simultaneous local metric generator for the family of lexicographic product graphs , then is a simultaneous local adjacency generator for .
In order to state our main result (Theorem 13), we need to introduce some additional notation. Let B be a simultaneous local adjacency generator for a family of non-trivial connected graphs on a common vertex set , and let be family of lexicographic product graphs defined as above.
.
If , then we define the graph in the following way. The vertex set of is , and two vertices are adjacent in if and only if for for every , .
If , then define , otherwise , where represents the domination number of .
is a family composed by empty graphs.
Notice that .
is the family of subsets of as follows. We say that if for every such that , for some , there exists such that .
is the graph with vertex set and edge set such that if and only if there exists such that .
Remark 8. if and only if for every .
Proof. If for every , then is the only simultaneous local adjacency basis of , , and then, . On the other hand, suppose that for some . In this case, . If , then we are done. Suppose that . For any simultaneous local adjacency basis of , there exists for some , which implies that and so . Therefore, , and the result follows. ☐
As we will show in the next example, in order to get the value of , it is interesting to remark about the necessity of considering the family of all simultaneous local adjacency generators and not just the family of simultaneous local adjacency bases of .
Example 1. Let be two copies of the path graph on five vertices such that , whereas and . Consider the family . We have that is a simultaneous local adjacency basis of and is a simultaneous local adjacency generator for . Then, , , , . However, and .
We define the following graph families.
is the family of graphs having at least two non-trivial components.
is the family of graphs having at least one component of radius at least four.
is the family of graphs having at least one component of girth at least seven.
is the family of graphs having at least two non-singleton true twin equivalence classes such that .
Lemma 3. Let be a family of graphs on a common vertex set V. If , then: Proof. Let B be a simultaneous local adjacency generator for and . We claim that . To see this, we differentiate the following cases for .
H has two non-trivial connected components . In this case, and , which implies that .
H has one non-trivial component J such that . If H has two non-trivial components, then we are in the first case. Therefore, we can assume that J is the only non-trivial component of H. Suppose that , and get such that . If is a shortest path from v to , then and are adjacent, and they are not distinguished by the elements in B, which is a contradiction.
H has one non-trivial component J of girth g. In this case, if H has two non-trivial components, then we are in the first case. Therefore, we can assume that H has just one non-trivial component of girth g. Suppose that . For each cycle , there exists such that and ; therefore, for each , we have and , which is a contradiction.
H has two non-singleton true twin equivalence classes such that . Since and , we can conclude that .
. Notice that , as , so that .
According to the five cases above, leads to , for any simultaneous local adjacency generator, which implies that . ☐
Remark 9. If , then is a simultaneous local metric generator for . However, the converse is not true, as we can see in the following example.
Example 2. Consider the family of connected graphs on a common vertex set with for . Let be the family consisting of only one graph , as follows: , . We have that and . If we take , then is a simultaneous local metric basis of . However, because is adjacent to in , , and .
Lemma 4. Let be a family of lexicographic product graphs. Let be a simultaneous local metric generator for . Then, .
Proof. Let and . Since is a simultaneous local metric generator for , for each , there exists such that . If , then necessarily , and if , then ; and we are done. ☐
Corollary 5. If there exists a simultaneous local metric generator B for such that or the graph is empty, then .
Remark 10. If B is a vertex cover of , then .
Lemma 5. Let be a family of lexicographic product graphs. For each , let be a simultaneous local adjacency generator for , and let be a dominating set of . Then, for any , the set is a local metric generator for .
Proof. In order to prove the lemma, let , , and let be a pair of adjacent vertices of . If , then there exists such that distinguishes the pair. Otherwise, , and we consider the following cases:
, say . In this case, there exists such that , and then, distinguishes the pair.
and
. In this case, by definition of
A, there exists
such that
. For any
,
and , say . In this case, if there exists such that , then distinguishes the pair. Otherwise, is a vertex of , and either and distinguishes the pair or there exists , such that , which means ; then, distinguishes the pair.
☐
Corollary 6. Let be a family of lexicographic product graphs. Then: Proof. Let
. For each
, let
be a simultaneous local adjacency basis of
. For each
, let
be a local adjacency generator for
and
a dominating set of
such that
. Let:
then, by Lemma 5,
B is a simultaneous local metric generator for
, and:
As
is arbitrary:
and the result follows. ☐
Lemma 6. Let F be a simultaneous local metric basis of . Let and . Then, .
Proof. Suppose, for the purpose of contradiction, that
. That means that there exist
and
such that
, and
for every
. As
,
and
, so that there exist
and
such that for some
,
,
and
. Let
be such that
. Consider the pair of vertices
adjacent in
. As
F is a simultaneous local metric generator, there exists
that resolves the pair, which implies that
. By hypothesis
, and so,
. Without loss of generality, we assume that
and, in this case,
which is a contradiction. Therefore,
. ☐
Theorem 13. Let be a family of lexicographic product graphs. Proof. Let
B be a simultaneous local metric basis of
. Let
and
. By Remark 7,
for every
, so that Lemma 6 leads to:
and the result follows by Corollary 6. ☐
Now, we will show some cases where the calculation of is easy. At first glance, we have two main types of simplification: first, to simplify the calculation of and, second, the calculation of the that makes the sum achieves its minimum.
For the first type of simplification, we can apply Lemma 3 to deduce the following corollary.
Corollary 7. If for each i, and , then: Given a family
of graphs on a common vertex set
V and a graph
H, we define the family of lexicographic product graphs:
By Theorem 13, we deduce the following result.
Corollary 8. Let be a family of graphs on a common vertex set V, and let H be a graph. If for every local adjacency basis B of H, for every , then: By Corollary 5 and Theorem 13, we have the following result.
Proposition 2. If is a simultaneous local metric generator for or the graph is empty, then: For the second type of simplification, we have the following remark.
Remark 11. As , if , then: From Remark 11, we can get some consequences of Theorem 13.
Proposition 3. Let be a family of lexicographic product graphs. For any vertex cover B of , Proposition 4. Let be a family of connected graphs with common vertex set V, and let be a family of lexicographic product graphs. The following statements hold.
- 1.
If the subgraph of induced by is empty for every , then: - 2.
Let be such that . If and , then:
Proof. It is clear that if the subgraph of induced by is empty for every , then , so that Theorem 13 leads to (1). On the other hand, let be a family of connected graphs with common vertex set V such that and . By Lemma 1, for every , there exists such that are true twins in . Hence, no vertex resolves and . Therefore, implies , and (2) follows from Theorem 13 and Remark 11. ☐
Proposition 5. Let be a family of non-trivial connected graphs with common vertex set V. For any family of lexicographic product graphs ,Furthermore, if , then: Proof. Let
W be a simultaneous local metric basis of
and
. We suppose that
is not a simultaneous local metric generator for
. Let
and
such that
and
for every
. Thus, for any
,
and
, we have:
which is a contradiction. Therefore,
is a simultaneous local metric generator for
and, as a result,
.
On the other hand, if , then . Let be a simultaneous local metric basis of . Now, for each , we choose , and by Remark 9, we claim that is a simultaneous local metric generator for . Thus, . ☐
Proposition 6. Let be a family of non-trivial connected bipartite graphs with common vertex set V and such that , for some j. If and there exist and such that and , then:otherwise, Proof. If and there exist and such that and , then because no vertex in distinguishes and . Let such that . Since any distinguishes x and y, we can conclude that , and by Remark 8, . Therefore, Theorem 13 leads to
Assume that there exists , or , or and, for every , or . In any one of these cases is a simultaneous local metric basis of and, for every pair of adjacent vertices in some such that , distinguishes the pair. Since , we can claim that , and by Theorem 13, ☐
5.1. Families of Join Graphs
For two graph families
and
, defined on common vertex sets
and
, respectively, such that
, we define the family:
Notice that, since for any
and
the graph
has diameter two,
The following result is a direct consequence of Theorem 13.
Corollary 9. For any pair of families and of non-trivial graphs on common vertex sets and , respectively, Remark 12. Let be a family of graphs defined on a common vertex set . If there exists B a simultaneous local adjacency basis of such that , then for every family of graphs defined on a common vertex set , we have: By Lemma 3 and Remark 12, we deduce the following result.
Proposition 7. Let and be two families of non-trivial connected graphs on a common vertex set and , respectively. If , then: 6. Computability of the Simultaneous Local Metric Dimension
In previous sections, we have seen that there is a large number of classes of graph families for which the simultaneous local metric dimension is well determined. This includes some cases of graph families whose simultaneous metric dimension is hard to compute, e.g., families composed by trees [
22], yet the simultaneous local metric dimension is constant. However, as proven in [
23], the problem of finding the local metric dimension of a graph is NP-hard in the general case, which trivially leads to the fact that finding the simultaneous local metric dimension of a graph family is also NP-hard in the general case.
Here, we will focus on a different aspect, namely that of showing that the requirement of simultaneity adds to the computational difficulty of the original problem. To that end, we will show that there exist families composed by graphs whose individual local metric dimensions are constant, yet it is hard to compute their simultaneous local metric dimension.
To begin with, we will formally define the decision problems associated with the computation of the local metric dimension of one graph and the simultaneous local metric dimension of a graph family.
Local metric Dimension (LDIM)
Instance: A graph and an integer p, .
Question: Is ?
Simultaneous Local metric Dimension (SLD)
Instance: A graph family on a common vertex set V and an integer p, .
Question: Is ?
As we mentioned above, LDIM was proven to be NP-complete in [
23]. Moreover, it is simple to see that determining whether a vertex set
,
, is a simultaneous local metric generator can be done in polynomial time, so SLD is in NP. In fact, SLD can be easily shown to be NP-complete, since for any graph
and any integer
, the corresponding instance of LDIM can be trivially transformed into an instance of SLD by making
.
For the remainder of this section, we will address the issue of the complexity added by the requirement of simultaneity. To this end, we will consider families composed by the so-called tadpole graphs [
28]. An
-tadpole graph (or
-tadpole for short) is the graph obtained from a cycle graph
and a path graph
by joining with an edge a leaf of
to an arbitrary vertex of
. We will use the notation
for
-tadpoles. Since
-tadpoles are bipartite, we have that
. In the case of
-tadpoles, we have that
, as they are not bipartite (so,
), and any set composed by two vertices of the subgraph
is a local metric generator (so,
). Additionally, consider the sole vertex
v of degree three in
and a local metric generator for
of the form
,
. It is simple to verify that for any vertex
, the set
is also a local metric generator for
.
Consider a family
composed by tadpole graphs on a common vertex set
V. By Theorem 4, we have that
, where
is composed by
-tadpoles. As we discussed previously,
. However, by Remark 1 and Theorem 1, we have that
. In fact, both bounds are tight, since the lower bound is trivially satisfied by unitary families, whereas the upper bound is reached, for instance, by any family composed by all different labeled graphs isomorphic to an arbitrary
-tadpole, as it satisfies the premises of Theorem 1. Moreover, as we will show, the problem of computing
is NP-hard, as its associated decision problem is NP-complete. We will do so by showing a transformation from the hitting set problem, which was shown to be NP-complete by Karp [
29]. The hitting set problem is defined as follows:
Hitting Set Problem (HSP)
Instance: A collection of non-empty subsets of a finite set S and a positive integer .
Question: Is there a subset with such that contains at least one element from each subset in ?
Theorem 14. The Simultaneous Local metric Dimension problem (SLD) is NP-complete for families of -tadpoles.
Proof. As we discussed previously, determining whether a vertex set , , is a simultaneous local metric generator for a graph family can be done in polynomial time, so SLD is in NP.
Now, we will show a polynomial time transformation of HSP into SLD. Let be a finite set, and let , where every satisfies . Let p be a positive integer such that . Let such that . We construct the family composed by -tadpoles on the common vertex set , , by performing one of the two following actions, as appropriate, for every :
If is even, let be a cycle graph on the vertices of ; let be a path graph on the vertices of ; and let be the tadpole graph obtained from and by joining with an edge a leaf of to a vertex of different from u.
If is odd, let be a cycle graph on the vertices of ; let be a path graph on the vertices of ; and let be the tadpole graph obtained from and by joining with an edge the vertex to a leaf of .
Figure 4 shows an example of this construction.
In order to prove the validity of this transformation, we claim that there exists a subset of cardinality that contains at least one element from each if and only if .
To prove this claim, first assume that there exists a set , which contains at least one element from each and satisfies . Recall that any set composed by two vertices of is a local metric generator for , so is a simultaneous local metric generator for . Thus, .
Now, assume that
, and let
W be a simultaneous local metric generator for
such that
. For every
, we have that
and
, so
for any
. As a consequence, if
, any set
,
, is also a simultaneous local metric generator for
, so we can assume that
. Moreover, applying an analogous reasoning for every set
such that
, we have that, firstly, there is at least one vertex
such that
and
, and secondly, there is at least one vertex
, which can be replaced by
. Then, the set:
is also a simultaneous local metric generator for
of cardinality
such that
and
for every
. Thus, the set
satisfies
and contains at least one element from each
.
To conclude our proof, it is simple to verify that the transformation of HSP into SLD described above can be done in polynomial time. ☐