1. Introduction
In chemical graph theory, the molecular structure of a compound is represented as a graph. In this context, atoms are represented as vertices, and chemical bonds are represented as edges connecting these vertices. One of the main goals of chemical graph theory is to analyze the molecular structure of a chemical compound through the study of its molecular graph, using graph theoretical and computation techniques [
1].
A topological index, also called a molecular descriptor, is a numerical parameter of a graph which is invariant under graph isomorphism. It plays an important role in chemical graph theory, especially in the quantitative structure–property relationship (QSPR) and the quantitative structure–activity relationship (QSAR) investigations [
2,
3]. There are a variety of topological indices that are derived from different concepts such as entropy [
4] and counting polynomials [
5]. One important class of topological indices are the vertex-degree-based topological indices, which are defined in terms of the degrees of the vertices of the graph.
More precisely, let
be a graph with vertex set
and edge set
. A vertex-degree-based (VDB for short) topological index
is defined for the graph
G as
where
denotes the degree of the vertex
, and
is an appropriate function with the property
. For recent results on VDB topological indices, we refer to [
6,
7,
8,
9], where extremal value problems in significant classes of graphs are solved. A geometrical approach to VDB topological indices is considered in [
10,
11], and in [
12], applications to COVID-19 are found.
Among the VDB topological indices, the general sum-connectivity indices are defined as
They were introduced by Zhou and Trinajstić [
13,
14] and have attracted the attention of many researchers due to their interesting mathematical properties and chemical applicability. For instance, in [
15], it is shown that there is a good correlation between physico-chemical properties and the general sum-connectivity index for benzenoid hydrocarbons. Bounds for the general sum-connectivity index in different classes of graphs are found in [
16,
17,
18,
19]. In [
20,
21], lower bounds of the line graph are found, and extremal graphs are characterized.
Another type of topological index is the well-known Hosoya index, denoted by
and introduced by Haruo Hosoya in 1971 [
22] to report a good correlation of the boiling points of alkane isomers. In order to properly define it, we must recall that a set of edges
is called a matching of
G if no two edges of
M have a vertex in common. A matching of
G with
k edges is said to be a
k-matching. We denote by
the number of
k-matchings of
G and assume that
. Then, the Hosoya index
is defined as
It is well known that if
T is a tree with
n vertices; then,
for all
, where
is the star on
n vertices, and
is the path on
n vertices (see [
23] (Theorem 4.6)).
In view of its structural relationship to both the VDB topological index
and the Hosoya topological index
Z, the construction of a novel molecular descriptor was introduced in [
24], called the Hosoya index of VDB-weighted graphs. It is defined as
where
, and for
,
It turns out that
so for
, we recover the VDB topological index
, and for
, the
k-matching numbers
can be viewed as
kth order topological indices which involve both the topological index
and the
k-matching numbers.
It is our main interest in this paper to study when T is a tree, and is the general sum-connectivity index. Concretely, we find the extremal values of the function , where is the set of trees with n vertices, and is a real number in the interval
2. 2-Matchings with Respect to
By
, we denote the
n-vertex star. It is straightforward that for any VDB topological index
and
different from the star,
Consequently, among all trees with
n vertices,
attains the minimum number of 2-matchings with respect to any VDB topological index
.
In the next example, we show how to compute the number of 2-matchings with respect to any VDB topological index
of the trees depicted in
Figure 1.
Example 1. Consider the tree depicted in Figure 1, and let be the set of 2-matchings of . The elements of are , , , , and . Then, the number of 2-matchings of with respect to φ is given by The elements of the set of 2-matchings of in Figure 1 are , , , , , , , , , , , , , , , , and . Consequently, For the tree of Figure 1, the elements of are , , , , , , , , , , , , , , , , , , , and . It follows that In this section, we consider 2-matchings with respect to , where , over the set of trees with vertices. In order to find the second minimum, we first analyze 2-matchings with respect to over the set of double-star trees.
Let and . The double-star is a tree with exactly two vertices of degree greater than 1, one having degree and the other one having degree . It is easy to see that .
We distinguish two extreme double-stars with respect to parameters
p and
q, the double star
and the balanced double-star
(see
Figure 2).
Lemma 1. Let and . If , then Proof. Let
defined for
. The derivative
since
for
and
for
and
. Consequently,
, and it is completed. □
Next, we compute the variation of 2-matchings with respect to
when “Transformation A” described in [
25] is performed. Let
,
and
where
and
are leaves. Let
and
(see
Figure 3).
Note that
. For any other vertex in
, we denote
. Let
, where
means that
and
are independent edges, and
Then,
where
.
Now, we can prove that the double-star attains the second minimum number of 2-matchings with respect to over the set .
Theorem 1. Let such that and . If , then, Proof. Assume
is not a double-star and is of the form depicted in
Figure 3; then,
is a subtree of
T different from a star with
v as a central vertex. If
is the tree obtained from
T by “Transformation A”, then,
is not a star. Since
, using relation (
1), we obtain
Applying this transformation repeatedly, we obtain a sequence of trees such that , and is a double-star with n vertices. Now, the result follows from Lemma 1. □
Using appropriate values of
in the previous theorem, we obtain the results for the sum-connectivity index
[
13] and the harmonic index
[
26,
27].
Corollary 1. Let and such that and .
- 1.
.
- 2.
.
Next, we show that for
, the maximum number of 2-matchings with respect to
is attained in the path
, among all trees in
. It is easy to see that
Recall that a branching vertex of a tree
T is a vertex of degree
. If
v is a branching vertex of degree
k of a tree
T, then
T can be viewed as the coalescence of
k subtrees of
T at the vertex
v. A branching vertex
v of
T is an outer branching vertex of
T if all branches of
T at
v (except for possibly one) are paths. The concept of the outer branching vertex was introduced in [
28]. In the mentioned paper, it was shown that a tree
has no outer branching vertex if and only if
.
Now, we compute the variation of 2-matchings with respect to
when “Transformation C” described in [
25] is performed.
Let
,
; then, there exists an outer branching vertex
and two paths
and
(branches at
) with
. Then,
T can be viewed as a coalescence at vertex
of a subtree
and the path
. We construct the tree
as
(see
Figure 4).
Let ; then, and . For any other vertex in , we denote . We also distinguish the subset of edges .
Now, we compute the difference , where . By symmetry, we may assume and consider only the following three cases:
Theorem 2. Let and such that . If , then, Proof. Since
with
, then
T has the form depicted in
Figure 4. Let
be a tree obtained from
T by “Transformation C”. For
, the expression
for any integer
. We have to consider the three cases in “Transformation C”.
For
and
, by relation (
2),
if
. The same occurs for
and
. By relation (
3),
if
.
For the case
and
, by relation (
4), we have to prove that
for
. Note that
Let
defined for
. The derivative
Note that for , and . It follows that for , which means that is strictly decreasing for . Then, Consequently, in this case, if .
Applying this transformation repeatedly, we obtain a sequence of trees such that and . □
Using appropriate values of in the previous theorem, we obtain the results for the sum-connectivity index and the harmonic index .
Corollary 2. Let and such that .
- 1.
.
- 2.
.
In the following example, we show that if is not in this interval, the result is not necessarily true.
Example 2. For , we have , the First Zagreb index [29], defined by . It is easy to see thatandfor . Consequently, the path is not an extreme tree with respect to the number of 2-matchings of trees weighted with the First Zagreb index. The same situation occurs with the Second Zagreb index
[
29] defined by
, as we can see in our next example.
Example 3. It is easy to check thatandfor . Hence, the path is not an extreme tree with respect to the number of 2-matchings of trees weighted with the Second Zagreb index. Finally, we give an example of a VDB topological index where is not an extreme tree over .
Example 4. In the case of the Forgotten index [30], defined by , we haveandfor . In other words, the double-star is not an extreme tree with respect to the number of 2-matchings of trees weighted with the Forgotten index. 3. Conclusions
The novel topological indices
of a graph
G with respect to a VDB topological index
described in [
24] are natural extensions of VDB topological indices, which involve both the topological index
and the
k-matching numbers, perhaps two of the most important concepts of chemical graph theory. In this paper, we initiate the study of 2-matchings with respect to general sum-connectivity indices over the significant class of trees with a fixed number of vertices. The techniques used here are successful in showing that the extremal values of the function
are attained in the path
and the double-star
, when
is a real number in the interval
As we noted in Example 2, the result is no longer true when
. So, a first natural question is the following:
Problem 1. Find the extremal values of , when
Other important types of vertex-degree-based topological indices are the general Randić indices
[
31,
32], which are obtained from the symmetric functions
where
. Using the same technique with some minor adaptations, we were able to show that the double-star
attains the second minimal value of
when
. However, we failed in showing that the path
attains the maximal value. So, another problem is the following:
Problem 2. Find the maximal value of , when
On the other hand, it would be of great interest to determine extremal values of or, more generally, of , when G belongs to other interesting classes of graphs, for instance, chemical trees, hexagonal systems or unicyclic graphs, just to mention a few.