Research Article
It has been observed that the graph of an 𝑘-gonal (where
𝑘∈N and 𝑘≥3) prism exhibits a distinct pattern in
terms of its vertices and edges. Specifically, it has been
determined that an 𝑘-gonal prism possesses 2𝑘 vertices
and 3𝑘 edges. This finding provides valuable insight
into the structural properties of 𝑘-gonal prisms and
contributes to our understanding of their geometric
characteristics. The graphs in question exhibit regularity
and possess a cubic structure. Prism graphs possess the
property of vertex-transitivity due to the presence of
symmetries that map each vertex to every other vertex. As
polyhedral graphs, they exhibit the property of being
3-vertex-connected planar graphs. It has been observed
that every prism graph, which is a specific type of graph
formed by connecting two copies of a cycle graph with
corresponding vertices, possesses a Hamiltonian cycle. A
Hamiltonian cycle is a cycle that visits each vertex
exactly once. Researchers have extensively studied this
property and proven its validity for all prism graphs
[1].
Throughout this article, all graphs under consideration are
connected, non-trivial, undirected, and simple. Let
Γ𝑟 =(𝑉,𝐸) be a graph with 𝑉(Γ𝑟) and 𝐸(Γ𝑟) as its
vertex and edge set respectively. The shortest length path
between two distinct vertices 𝑦1 and 𝑦2 in 𝑉(Γ𝑟), is
referred to as the distance (𝑑(𝑦1,𝑦2)) between 𝑦1 and 𝑦2
in Γ𝑟. The number of distinct edges incident on a vertex 𝑦
in Γ𝑟 is known as the degree of 𝑣 (denoted by 𝑑𝑦).
Two vertices 𝑦1 and 𝑦2 in Γ𝑟 are said to resolved
by a vertex 𝑦, if 𝑑(𝑦,𝑦1)≠ 𝑑(𝑦,𝑦2) in Γ𝑟. Then,
a subset 𝑈⊆𝑉(Γ𝑟) with this property, i.e., every
pair of unequal vertices in Γ𝑟 can be resolved by
at least one member of 𝑈, is said to be a resolving
set (RS) for Γ𝑟. The smallest cardinality set 𝑈 with
resolving characteristic is called the metric basis
(MB) for Γ𝑟, and the MB set cardinality is the metric
dimension (MD) for Γ𝑟, represented by 𝑑𝑖𝑚(Γ𝑟)
[2, 3].
For a subset 𝑈={𝑦1,𝑦2,𝑦3,...,𝑦𝑠} of distinct ordered vertices in 𝑉(Γ𝑟), the unique 𝑠-length tuple code for each 𝑞∈𝑉(Γ𝑟) is given as follows ζ(𝑞|𝑈)=(𝑑(𝑞,𝑦1),𝑑(𝑞,𝑦2),𝑑(𝑞,𝑦3),...,𝑑(𝑞,𝑦𝑠)). Using this fact, the subset 𝑈 is a RS for Γ𝑟, if ζ(𝑦1|𝑈)≠ ζ(𝑦2|𝑈), for every pair of different vertices 𝑦1,𝑦2 ∈𝑉(Γ𝑟). Next, a subset 𝑈 in 𝑉(Γ𝑟) with distinct vertices is said to be a resolving independent set for Γ𝑟, if it is (i) independent set as well as (ii) a RS in Γ𝑟. A proper subset of a RS is not necessarily a RS, while a superset of every RS is always aRS [4].
To understand the concept of RS and MD, let us consider a
graph Γ* on 5 vertices and 7 edges, as shown in Fig. 1. To
find the metric dimension of Γ*, we suppose that
𝑈1 ={𝑣1,𝑣4} (red color vertices in Γ*). Next, metric codes
for each vertex in Γ* with respect to 𝑈1 are as follows:
ζ(𝑣1|𝑈1)=(0,1), ζ(𝑣2|𝑈1)=(1,2), ζ(𝑣3|𝑈1)=(2,2),
ζ(𝑣4|𝑈1)=(1,0), and ζ(𝑣5|𝑈1)=(1,1). From this, we
find that the metric codes for all the vertices in Γ*
corresponding to the set 𝑈1 are unique, and so we say that
𝑈1 is a resolving set for Γ*. Also, 𝑈1 is the minimum
resolving set for Γ*, as the cardinality of 𝑈1 is 2 [5].
Hence, we have concluded that 𝑑𝑖𝑚(Γ*)=2.
Slater [3] and Harary & Melter [2] independently
introduced the concept of MD of a graph in 1970s.
There is a wide range of literature available on MD
that addresses both theoretical and practical aspects.
The MD has appeared in various areas including
combinatorial optimization, sonar, pharmaceutical
chemistry, robot navigation, graph isomorphism testing,
and many more see [6, 7, 8, 9, 10] and references
therein.
The computation of MD for distinct graph families, is
always a challenging task because deciding and selecting
of a landmark (resolving) vertices in minimum numbers is
not that easy due to the complexity and scalability of the
considered graph network. Further, the notion of MD was
extended as well as investigated by eminent researchers
from time to time and named them as the variants of
MD [11]. Several authors have studied MD and its
related variants for several distinct graphs as well as for
various other graph-theoretic aspects, for instance prism
graph, path graph, complete graph, cycle graph, cycle
graph with chords, antiprism graph, several ladder
graphs (pentagonal, heptagonal, etc), convex polytope
graph, wheel graph, tadpole graph, kayak paddle
graph, and numerous planar and chemical graphs
[4, 6, 7, 9, 10, 12, 13]. Even though after investigating
these notions for the large number of graph families, there
are still many families for which these notions are not
investigated to date.
In this paper, two planar graph families, viz., L𝑛 and M𝑛
has been constructed, which are obtained by taking
𝑚-copies (𝑞= 𝑚) of the prism graph on the two
known convex polytope graphs N𝑛 [9] and 𝑈𝑛 [7],
respectively. For these so obtained graphs, we investigate
their minimal MB sets and finally obtain their MD.
To carry out these results, we need the following
result:
Proposition 1.1. [5] For a connected graph Γ𝑟 with metric dimension two i.e., the metric basis 𝑈for Γ𝑟 has cardinality two, and say 𝑈={𝑦1,𝑦2}. Then, the following three points for Γ𝑟 are true:
Always shortest unique path 𝑃between 𝑦1 and 𝑦2 exist,
𝑑𝑦1 and 𝑑𝑦2 is at most 3, and
The 𝑑𝑣 is at most 5, for any 𝑣∈𝑃.
The following paper is structured as follows: In Section 2, we consider an infinite family of convex polytope L𝑛 and find its minimum RS with its respective MD. In Section 3, we consider an infinite family of convex polytope M𝑛 and find its minimum RS with its respective MD. Finally, the conclusion and future scope of the manuscript is presented.
In this section, we investigate some of the basic properties
and the MD of a planar graph L𝑛. The graph L𝑛 consists
of 𝑛(𝑞+3) vertices and 𝑛(2𝑞+5) edges (see Fig.2). The
sets containing vertices and edges for planar graph L𝑛 are
denoted by 𝑉(L𝑛) and 𝐸(L𝑛) respectively, where
𝐸(L𝑛)={𝑗α 𝑗α+1 ,𝑗α 𝑘α ,𝑘α 𝑗α+1 ,𝑘α 𝑙α ,𝑙α 𝑚α 1,𝑚α 1𝑙α+1 ,𝑚α 𝑠𝑚α+1 𝑠 : 1 ≤α≤𝑛,1 ≤𝑠≤𝑞}∪{𝑚α 𝑠𝑚α 𝑠+1 : 1 ≤α≤𝑛,1 ≤𝑠≤𝑞-1}
and 𝑉(L𝑛)={𝑗α ,𝑘α ,𝑙α ,𝑚α 𝑠 : 1 ≤α≤𝑛,1 ≤𝑠≤𝑞}.
We call vertices {𝑗α : 1 ≤α≤𝑛}, as 𝑗-cycle vertices in L𝑛, the vertices {𝑘α ,𝑙α : 1 ≤α≤𝑛}, as inner vertices in L𝑛, and the vertices {𝑚α 𝑠 : 1 ≤α≤𝑛,1 ≤𝑠≤𝑞}, as outer vertices in L𝑛. In the following result, we investigate the MD of L𝑛.
Proof. Now, the following cases, which depend on the
natural 𝑛, can be employed to investigate this result.
Case(I) 𝑛≡0(𝑚𝑜𝑑 2).
We set 𝑛=2𝑦; 𝑦∈Z+ and 𝑦≥3. Let 𝑈={𝑗2,𝑗𝑦+1,𝑗𝑛}⊂𝑉(L𝑛).
Next, each vertex of L𝑛 has given metric coordinate
corresponding to the taken set 𝑈.
For vertices over 𝑗-cycle, i.e., {𝑗α : 1 ≤α≤𝑛}, the
metric co-ordinates are
For the vertices {𝑘α : 1 ≤α≤𝑛}, the metric co-ordinates are
For the vertices {𝑙α : 1 ≤α≤𝑛}, the metric co-ordinates are ζ(𝑙α |𝑈)= ζ(𝑘α |𝑈)+(1,1,1) for 1 ≤α≤𝑛. Next, for the vertices {𝑚α 1 : 1 ≤α≤𝑛}, the metric co-ordinates are
![]() |
Now, for reverse inequality i.e., 𝑑𝑖𝑚(L𝑛)≥3, we show that no set 𝑈 with |𝑈|=2, form a RS for L𝑛. Assuming 𝑑𝑖𝑚(L𝑛)=2 (on contrary). Using proposition 1, we have following to discuss for RS 𝑈 with |𝑈|= 2 in L𝑛:
Let 𝑈= {𝑘1,𝑘𝑔}, 𝑘𝑔 (2 ≤ 𝑔≤ 𝑦+1), then ζ(𝑗𝑛|𝑈) = ζ(𝑘𝑛|𝑈), for 2 ≤ 𝑔≤ 𝑦-1, ζ(𝑙2|𝑈) = ζ(𝑘𝑛-1|𝑈), when 𝑔= 𝑦, and ζ(𝑗2|𝑈) = ζ(𝑗1|𝑈), when 𝑔= 𝑦+ 1, a contradiction.
Let 𝑈= {𝑙1,𝑙𝑔}, 𝑙𝑔 (2 ≤ 𝑔≤ 𝑦+1), then ζ(𝑗𝑛|𝑈) = ζ(𝑘𝑛|𝑈), for 2 ≤ 𝑔≤ 𝑦-1, ζ(𝑝3|𝑈) = ζ(𝑚22|𝑈), when 𝑔= 𝑦, and ζ(𝑗2|𝑈) = ζ(𝑗1|𝑈), when 𝑔= 𝑦+ 1, a contradiction.
Let 𝑈= {𝑚1𝑞,𝑚𝑔𝑞}, 𝑚𝑔𝑞 (2 ≤ 𝑔≤ 𝑦+1), then ζ(𝑚1𝑞-1|𝑈)= ζ(𝑚𝑛𝑞|𝑈), for 2 ≤𝑔≤𝑦, and ζ(𝑚2𝑞|𝑈)= ζ(𝑚𝑛𝑞|𝑈), when 𝑔= 𝑦+1, a contradiction.
Let 𝑈= {𝑘1,𝑙𝑔}, 𝑙𝑔 (1 ≤ 𝑔≤ 𝑦+1), then ζ(𝑗𝑛|𝑈) = ζ(𝑘𝑛|𝑈), for 1 ≤ 𝑔≤ 𝑦-1, ζ(𝑚11|𝑈) = ζ(𝑝3|𝑈), when 𝑔= 𝑦, and ζ(𝑗2|𝑈) = ζ(𝑗1|𝑈), when 𝑔= 𝑦+ 1, a contradiction.
Let 𝑈= {𝑘1,𝑚𝑔𝑞}, 𝑚𝑔𝑞 (1 ≤𝑔≤𝑦+1), then ζ(𝑗1|𝑈) = ζ(𝑗2|𝑈), for 𝑔= 1, ζ(𝑚𝑛1|𝑈) = ζ(𝑘2|𝑈), when 2 ≤𝑔≤𝑦, and ζ(𝑚𝑦+11|𝑈)= ζ(𝑙𝑦+2|𝑈), when 𝑔=𝑦+1, a contradiction.
Let 𝑈= {𝑙1,𝑚𝑔𝑞}, 𝑚𝑔𝑞 (1 ≤𝑔≤𝑦+1), then ζ(𝑗1|𝑈) = ζ(𝑗2|𝑈), for 𝑔= 1, ζ(𝑚𝑛-11|𝑈) = ζ(𝑗2|𝑈), when 2 ≤ 𝑔≤ 𝑦-1, ζ(𝑚𝑦1|𝑈) = ζ(𝑚𝑦-12|𝑈), when 𝑔= 𝑦, and ζ(𝑚𝑦+11|𝑈) = ζ(𝑚𝑦+22|𝑈), when 𝑔=𝑦+1, a contradiction.
Thus, from this we have 𝑑𝑖𝑚(L𝑛)≥3, implying that
𝑑𝑖𝑚(L𝑛)=3, ∀𝑛≥6.
Case(II) 𝑛≡1(𝑚𝑜𝑑 2).
We set 𝑛= 2𝑦+1; 𝑦∈Z+ and 𝑦≥3. Let
𝑈={𝑗2,𝑗𝑦+1,𝑗𝑛}⊂𝑉(L𝑛). Next, each vertex of L𝑛 has
given metric coordinate corresponding to the taken set 𝑈.
For vertices over 𝑗-cycle, i.e., {𝑗α : 1 ≤α≤𝑛}, the metric co-ordinates are
For the vertices {𝑘α : 1 ≤α≤𝑛}, the metric co-ordinates are
Forthe vertices {𝑙α : 1 ≤α≤𝑛}, the metric co-ordinates are ζ(𝑙α |𝑈)=ζ(𝑘α |𝑈)+(1,1,1) for 1 ≤α≤𝑛. Next, for the vertices {𝑚α 1 : 1 ≤α≤𝑛}, the co-ordinates are
Finally, for the vertices {𝑚α 𝑠 : 1 ≤α≤𝑛,2 ≤𝑠≤𝑞}, the metric co-ordinates are ζ(𝑚α 𝑠|𝑈)=ζ(𝑚α 1|𝑈)+(𝑠-1,𝑠-1,𝑠-1) for 1 ≤α≤𝑛. Next, these codes for all vertices in L𝑛 are unique and distinct from one and an other in at least one co-ordinate, which results in 𝑑𝑖𝑚(L𝑛)≤3. Assuming that 𝑑𝑖𝑚(L𝑛)= 2, then as in Case (I), we have the same contradictions. Therefore, we have 𝑑𝑖𝑚(L𝑛)= 3 as well in this case, which proofs the theorem. □
Proof. For the proof, follow Theorem 2.1. □
In this section, we investigate some of the basic properties
and the MD of a planar graph M𝑛 (see Fig. 3). The graph
M𝑛 consists of 𝑛(𝑞+4) vertices and 𝑛(2𝑞+6) edges. The
sets containing vertices and edges for planar graph M𝑛 are
denoted by 𝑉(M𝑛) and 𝐸(M𝑛) respectively, where 𝐸(M𝑛)={𝑗α 𝑗α+1 ,𝑗α 𝑘α ,𝑘α 𝑘α+1 ,𝑘α 𝑙α ,𝑙α 𝑚α ,𝑚α 𝑙α+1 ,𝑚α 𝑜α 1,𝑜α 𝑠𝑜α+1 𝑠 : 1 ≤α≤𝑛,1 ≤𝑠≤
𝑞}∪{𝑜α 𝑠𝑜α 𝑠+1 : 1 ≤α≤𝑛,1 ≤𝑠≤𝑞-1} and
𝑉(M𝑛)={𝑗α ,𝑘α ,𝑙α ,𝑚α ,𝑜α 𝑠 : 1 ≤α≤𝑛,1 ≤𝑠≤𝑞}.
We call vertices {𝑗α : 1 ≤α≤𝑛}, as 𝑗-cycle vertices in
M𝑛, the vertices {𝑘α : 1 ≤α≤𝑛}, as 𝑘-cycle vertices in
M𝑛, the vertices {𝑙α ,𝑚α : 1 ≤α≤𝑛}, as 𝑙𝑚-cycle vertices
in M𝑛, and the vertices {𝑜α 𝑠 : 1 ≤α≤𝑛,1 ≤𝑠≤𝑞}
as outer vertices in M𝑛. In the following result, we
investigate the MD of M𝑛.
Proof. Now, the following cases, which depend on
the natural 𝑛, can be employed to investigate this
result.
Case(I) 𝑛≡0(𝑚𝑜𝑑 2).
We set 𝑛= 2𝑦; 𝑦∈ Z+ and 𝑦≥ 3. Let
𝑈={𝑗2,𝑗𝑦+1,𝑗𝑛}⊂𝑉(M𝑛). Next, each vertex of M𝑛 has
given metric coordinate corresponding to the taken set 𝑈.
For vertices over 𝑗-cycle, i.e., {𝑗α : 1 ≤α≤𝑛}, the
metric co-ordinates are
For the vertices {𝑘α : 1 ≤α≤𝑛}, the metric co-ordinates
are ζ(𝑘α |𝑈)= ζ(𝑗α |𝑈)+(1,1,1) for 1 ≤α≤𝑛.
Next, for the vertices {𝑙α : 1 ≤α≤𝑛}, the metric
co-ordinates are ζ(𝑙α |𝑈)= ζ(𝑘α |𝑈)+(1,1,1) for
1 ≤α≤𝑛.
For the vertices {𝑚α : 1 ≤α≤𝑛}, the metric co-ordinates
are
Finally, for the vertices {𝑜α 𝑠 : 1 ≤α≤𝑛,1 ≤𝑠≤𝑞}, the
metric co-ordinates are ζ(𝑜α 𝑠|𝑈)=ζ(𝑚α |𝑈)+(𝑠,𝑠,𝑠)
for 1 ≤α≤𝑛.
Next, these codes for all vertices in M𝑛 are unique and distinct from one and an other in at least one co-ordinate, which results in 𝑑𝑖𝑚(M𝑛)≤3. Now, for reverse inequality i.e., 𝑑𝑖𝑚(M𝑛)≥3, we show that no set 𝑈 with |𝑈|= 2, form a RS for M𝑛. Now, for reverse inequality i.e., 𝑑𝑖𝑚(M𝑛)≥3, we show that no set 𝑈 with |𝑈|= 2, form a RS for M𝑛. Assuming 𝑑𝑖𝑚(M𝑛)=2 (on contrary). Using proposition 1, we have following to discuss for RS 𝑈 with |𝑈|= 2 in M𝑛:
Let 𝑈= {𝑘1,𝑘𝑔}, 𝑘𝑔 (2 ≤ 𝑔≤ 𝑦+1), then ζ(𝑗𝑛|R) = ζ(𝑘𝑛|R), for 2 ≤ 𝑔≤ 𝑦-1; ζ(𝑙2|R) = ζ(𝑘𝑛-1|R), when 𝑔= 𝑦; and ζ(𝑗2|R) = ζ(𝑗1|R), when 𝑔= 𝑦+1, a contradiction.
Let 𝑈= {𝑙1,𝑙𝑔}, 𝑙𝑔 (2 ≤ 𝑔≤ 𝑦+1), then ζ(𝑗𝑛|R) = ζ(𝑘𝑛|R), for 2 ≤ 𝑔≤ 𝑦-1; ζ(𝑝3|R) = ζ(𝑚22|R), when 𝑔= 𝑦; and ζ(𝑗2|R) = ζ(𝑗1|R), when 𝑔= 𝑦+ 1, a contradiction.
Let 𝑈= {𝑚1𝑞,𝑚𝑔𝑞}, 𝑚𝑔𝑞 (2 ≤ 𝑔≤ 𝑦+1), then ζ(𝑚1𝑞-1|R)= ζ(𝑚𝑛𝑞|R), for 2 ≤𝑔≤𝑦; and ζ(𝑚2𝑞|R)= ζ(𝑚𝑛𝑞|R), when 𝑔= 𝑦+1, a contradiction.
Let 𝑈= {𝑘1,𝑙𝑔}, 𝑙𝑔 (1 ≤ 𝑔≤ 𝑦+ 1), then ζ(𝑗𝑛|R) = ζ(𝑘𝑛|R), for 1 ≤ 𝑔≤ 𝑦-1; ζ(𝑚11|R) = ζ(𝑝3|R), when 𝑔= 𝑦; and ζ(𝑗2|R) = ζ(𝑗1|R), when 𝑔= 𝑦+1, a contradiction.
Let 𝑈= {𝑘1,𝑚𝑔𝑞}, 𝑚𝑔𝑞 (1 ≤ 𝑔≤ 𝑦+1), then ζ(𝑗1|R)=ζ(𝑗2|R), for 𝑔=1; ζ(𝑚𝑛1|R)= ζ(𝑘2|R), when 2 ≤ 𝑔≤ 𝑦; and ζ(𝑚𝑦+11|R) = ζ(𝑙𝑦+2|R), when 𝑔=𝑦+1, a contradiction.
Let 𝑈= R = {𝑙1,𝑚𝑔𝑞}, 𝑚𝑔𝑞 (1 ≤ 𝑔≤ 𝑦+1), then ζ(𝑗1|R) = ζ(𝑗2|R), for 𝑔= 1; ζ(𝑚𝑛-11|R) = ζ(𝑗2|R), when 2 ≤ 𝑔≤ 𝑦-1; ζ(𝑚𝑦1|R) = ζ(𝑚𝑦-12|R), when 𝑔= 𝑦; and ζ(𝑚𝑦+11|R) = ζ(𝑚𝑦+22|R), when 𝑔= 𝑦+1, a contradiction.
Thus, from this we have 𝑑𝑖𝑚(M𝑛)≥3, implying that
𝑑𝑖𝑚(M𝑛)=3, ∀𝑛≥6.
Case(II) 𝑛≡1(𝑚𝑜𝑑 2).
We set 𝑛= 2𝑦+1, 𝑦∈ Z+ and 𝑦≥ 3. Let
𝑈={𝑗2,𝑗𝑦+1,𝑗𝑛}⊂𝑉(M𝑛). Next, each vertex of M𝑛 has
given metric coordinate corresponding to the taken set 𝑈.
For vertices over 𝑗-cycle, i.e., {𝑗α : 1 ≤α≤𝑛}, the
metric co-ordinates are
For the vertices {𝑘α : 1 ≤α≤𝑛}, the metric co-ordinates
are ζ(𝑘α |𝑈)= ζ(𝑗α |𝑈)+(1,1,1) for 1 ≤α≤𝑛.
Next, for the vertices {𝑙α : 1 ≤α≤𝑛}, the metric
co-ordinates are ζ(𝑙α |𝑈)= ζ(𝑘α |𝑈)+(1,1,1) for
1 ≤α≤𝑛.
For the vertices {𝑚α : 1 ≤α≤𝑛}, the metric co-ordinates are
Finally, for the vertices {𝑜α 𝑠 : 1 ≤α≤𝑛,1 ≤𝑠≤𝑞}, the
metric co-ordinates are ζ(𝑜α 𝑠|𝑈)=ζ(𝑚α |𝑈)+(𝑠,𝑠,𝑠)
for 1 ≤α≤𝑛.
Next, these codes for all vertices in M𝑛 are unique
and distinct from one and an other in at least one
co-ordinate, which results in 𝑑𝑖𝑚(M𝑛)≤3. Assuming
that 𝑑𝑖𝑚(M𝑛)=2, then as in Case (I), we have the same
contradictions. Therefore, we have 𝑑𝑖𝑚(M𝑛)=3 as well
in this case, which proves the theorem.
□
Proof. For the proof, follow Theorem 3.1. □
Obtaining resolving set for novel planar structure plays an important role in interconnection networks for the transmission of the data. In this article, we proved that 𝑑𝑖𝑚(L𝑛)=𝑑𝑖𝑚(M𝑛)=3, where L𝑛 and M𝑛 are obtained by joining 𝑞-copies of the prism graph on N𝑛 [9] and 𝑈𝑛 [7], respectively. Further, we demonstrated for these two planar convex polytope graphs that the cardinality of respective resolving independent set is also three for them. Further, several other variations of MD were also introduced in last two decades, such as edge metric dimension, local metric dimension, strong metric dimension, mixed metric dimension, etc [11, 12, 14]. Therefore, in future we will try to investigate several other variants of MD for the planar graphs L𝑛 and M𝑛.
We would like to express our sincere gratitude to the referees for their careful reading of this manuscript and for all of their insightful comments/criticism, which have resulted in a number of major improvements to this manuscript..
M. Gayathri is sole author of this article.
No funding was used in this study.
No conflicts of interest.
This research has been conducted ethically, reporting of those involved in this article.
I hereby confirm that there is no similarity index in abstract and conclusion while overall is less than 5% where individual source contribution is 2% or less than it.
Data sharing is not applicable to this article as no data set were generated or analyzed during the current study.
[1] M. A. Lisitsyna and S. V. Avgustinovich, Siberian Electron Math Rep Vol.13, 1116 (2016). https://doi.org/10.17377/semi.2016.13.088
[2] F. Harary and R. A. Melter, On the metric dimension of a graph, Ars Comb Vol.2, 191 (1976).
[3] P. J. Slater, Leaves of trees, Cong Numeran Vol.14, 549 (1975).
[4] G. Chartrand et al., Math Bohem Vol. 128, 379 (2003). http://dx.doi.org/10.21136/MB.2003.134003
[5] S. Khuller et al., Discret Appl Math Vol.70, 217 (1996). https://doi.org/10.1016/0166-218X(95)00106-2
[6] J. Cáceres et al., SIAM J Discrete Math Vol. 21(2), 423 (2007). https://doi.org/10.1137/050641867
[7] M. Imran et al., Comput Math with Appl Vol.60(9), 2629 (2010). https://doi.org/10.1016/j.camwa.2010.08.090
[8] A. Sebo and E. Tannier, Math Oper Res Vol.29(2), 383 (2004). https://doi.org/10.1287/moor.1030.0070
[9] M. A. Malik and M. Sarwar, Afr Math Vol.27, 229 (2016). https://doi.org/10.1007/s13370-015-0336-5
[10] S. K. Sharma and V. K. Bhat, Discrete Math Algorithms Appl Vol.13(1), 2050095 (2021). https://doi.org/10.1142/S1793830920500950
[11] S. K. Sharma et al., Chem Zvesti Vol.76(7), 4115 (2022). https://doi.org/10.1007/s11696-022-02151-x
[12] A. Kelenc et al., Discret Appl Math Vol.31, 204 (2018). https://doi.org/10.1016/j.dam.2018.05.052
[13] S. K. Sharma et al., Front Phys Vol.9, 600 (2021). https://doi.org/10.3389/fphy.2021.749166.
[14] A. Kelenc et al., Appl Math Comput Vol.314, 429 (2017). https://doi.org/10.1016/j.amc.2017.07.027
[© 2024 M. Gayathri] This is an Open Access article published in "Graduate Journal of Interdisciplinary
Research, Reports & Reviews" (Grad.J.InteR3) ISSN(E): 2584-2919 by Vyom Hans Publications. It is
published with a Creative Commons Attribution - CC-BY4.0 International License. This license permits
unrestricted use, distribution, and reproduction in any medium, provided the original author and source are
credited.
____________________________________________________________________________________________