On Peripheral Wiener Index Line Graphs
On Peripheral Wiener Index Line Graphs
83 (2020) 129-141
Communications in Mathematical
and in Computer Chemistry ISSN 0340 - 6253
Abstract
The peripheral Wiener index, P W (G) is the sum of the distances of all pairs of
vertices in the periphery of a graph G. In this paper it is shown that, an arbitrary
graph G is an induced subgraph of a graph H for which P W (H) = P W (L(H))
holds, where L(K) is the line graph of K. Using Pell-like equations, infinite families
of graphs G are constructed for which P W (G) = P W (L(G)) holds. A connection
between the peripheral Wiener index and the Zagreb index for graphs of small
diameter is presented. It is also demonstrated that the partition distance approach
applicable to the peripheral Wiener index, making some earlier results as very
special cases of this approach.
1 Introduction
Because of the seminal Wiener’s discovery of a close relation between boiling points of
certain alkanes and the Wiener index [32], it became perceptible that graph invariants,
also addressed to as topological indices, can be used to predict properties of chemical
-130-
compounds. During the related history, many additional topological indices have been
studied, in particular several Wiener-like indices such as the hyper-Wiener index [21], the
reciprocal Wiener index [16], and the terminal Wiener index [11, 27], to name just a few
of them.
Unless stated otherwise, the graphs considered in this paper are finite and connected.
If G is a graph, then m(G) denotes the size of G and n(G) the order of G. Let dG (u, v)
denote the usual shortest path distance between vertices u and v in G. The eccentricity
eG (v) or e(v) of a vertex v is the distance to a furthest vertex from v in G. The diameter,
diam(G) of G is the maximum eccentricity of G. A vertex v with eG (v) = diam(G) is
called a peripheral vertex of G. The set of peripheral vertices of G is called the periphery
of G and denoted by P (G). We will also use the notation p(G) = |P (G)| for the order of
the periphery. We note in passing that the periphery of a graph has many applications
(see eg. [19, 20]), in particular because (as stated in [4]), the peripheral attachments of a
molecule are quite often the deciding factor in chemical reactions; cf. also [33].
In [24] another Wiener-like topological index was introduced as follows. If G is a
graph, then its peripheral Wiener index P W (G) is the sum of the distances between all
pairs of peripheral vertices of G, that is,
X
P W (G) = d(u, v) .
{u,v}⊆P (G)
The peripheral Wiener index received an immediate attention. In [5] the main focus is
on the peripheral Wiener index of trees, while in [15] different upper and lower bounds
are proved. Moreover, sharp bounds on the difference between the Wiener index and the
peripheral Wiener index are derived. In this paper we continue the study of the peripheral
Wiener index with two main goals: to understand its behaviour on line graphs and to
demonstrate how the cut method is applicable to the peripheral Wiener index.
Recall that the line graph L(G) of a graph G is the graph with V (L(G)) = E(G),
two vertices of L(G) being adjacent if and only if the corresponding edges of G are
adjacent. The concept of line graph has found various applications in chemical research,
cf. [6,10,12,25]. In particular, iterated line graphs turned out to be chemically important,
see [7, 13, 30]. In this paper we are interested in cycle containing graphs G with the
property
The related problem has been extensively studied for the Wiener index, a survey of these
developments is given in [23].
Constructions of unicyclic and bicyclic graphs for which one of P W (G) > P W (L(G)),
P W (G) < P W (L(G)), or P W (G) = P W (L(G)) holds can be seen in [29]. In Sec-
tion 2 we show that any graph G is an induced subgraph of a graph H, such that
P W (H) = P W (L(H)) holds. In Section 3 we give two infinite families of graphs G
for which P W (G) = P W (L(G)) hold. Pell-like Diophantine equations are essential for
their analysis. Then, in Section 4, we give a connection between the peripheral Wiener in-
dex and the Zagreb index for graphs of small diameter. In Section 5 we point out that the
very general partition distance approach from [18] is applicable to the peripheral Wiener
index, making several earlier results a very special cases of this approach. A couple of
future research directions are indicated at the end.
Theorem 2.1. If G is a graph, then there exists a graph H such that G is an induced
subgraph of H and P W (H) = P W (L(H)). Moreover, H − V (G) is a bicyclic graph.
Proof. Let Hp,q , p, q, ≥ 2, be the graph obtained from disjoint cycles C2p+1 and C2q by
identifying a vertex of C2p+1 with a vertex of C2q . Setting V (C2p+1 ) = {u0 , u1 , . . . , u2p }
and V (C2q ) = {w0 , w1 , . . . , u2q−1 }, we may assume without loss of generality that in Hp,q
the vertex u0 is identified with w0 , see Fig. 1.
up
C2p+1 u0 w0 C2q wq
up+1
Let now H be the graph obtained from the disjoint union of G and Hp,q by adding
all possible edges between the vertices of G and the vertices u0 = w0 , u2p , and w2q−1 . In
-132-
other words, in H there is a join between G and the subgraph of Hp,q induced by the
vertices u0 = w0 , u2p , and w2q−1 . See Fig. 2 where H is shown (left) and L(H) (right).
up e2 f2
fq
C2p+1 u0 w 0 C2q wq ep L(C2p+1 ) e1 f1 L(C2q )
up+1 fq+1
u2p w2q−1 e2p+1 f2q
L(G)
Selecting G to be the empty graph in Therem 2.1, we infer that the graphs Hp,q (as
defined in the proof of the theorem) form an infinite family of bicyclic graphs for which (1)
holds.
s µ−1 r
s µ r
v1 v2
a b
G(r, s, µ) H(r, s, µ)
Kµ−1 Kµ−1 Kµ Kµ
Ks Kr Ks Kr
Figure 3. Graphs G(r, s, µ), H(r, s, µ), and their line graphs.
To simplify the notation, set G = G(r, s, µ). Note that the order of G is µ + s + r + 3
and that p(G) = s + r + 2. Moreover, the order of L(G) is 2µ + s + r + 2 and L(G) is a
self-centered graph, that is, every vertex of it lies in the periphery. Hence
P W (G) = s2 + r2 + 3sr + 4r + 4s + 1 ,
1
6sµ + 6rµ + 4sr + s2 + r2 + 5s + 5r + 6µ2 + 8µ + 2 .
P W (L(G)) =
2
If the equality holds in (1) for G, then we get the following quadratic equation with respect
to the parameters s, r and µ:
The number s is an integer if and only if the radicand in (2) is an odd perfect square,
that is, y 2 = 60µ2 − 4µ + 9 for some odd integer y. Applying substitution x = 30µ − 1,
equality (2) can be rewritten as:
x + 5y 7
s+r = − , (3)
10 5
with the relation
15y 2 − x2 = 134 , (4)
which is a Pell-like Diophantine equation, see [28]. It has an infinite number of solutions.
All solutions of equation (4) can be generated by the following explicit formulae for integers
n ≥ 1:
(
xn = 4xn−1 + 15yn−1
(5)
yn = xn−1 + 4yn−1
√ √ √ √
with the initial values x0 + y0 15 = 1 + 3 15 or x0 + y0 15 = 41 + 11 15, since both
initial values are fundamental solutions of (4).
Consider next graphs H(r, s, µ) and L(H(r, s, µ)) as shown in Fig. 3. We again use
the parameter µ to indicate that the cyclomatic number of H(r, s, µ) is µ.
To simplify the presentation, set for the rest H = H(r, s, µ). The graph H has
µ + s + r + 2 vertices among which s + r are peripheral. So we have
P W (H) = 3sr + s2 + r2 − s − r .
1
P W (L(H)) = (6µs + 6µr + 4rs + s2 + r2 − s − r + 6µ2 − 4µ) .
2
If the equality holds in (1) for H, then
1
s2 + 3sr − s + r2 − r = (6µs + 6µr + 4rs + s2 + r2 − s − r + 6µ2 − 4µ).
2
Solving this for s, we have a suitable solution of the form
p
6µ − 2r + 1 + 60µ2 − 4µ + 1
s= . (6)
2
The number s is an integer if and only if y 2 = 60µ2 − 4µ + 1, for some odd integer y.
Using substitution x = 30µ − 1, equality (6) can be presented as:
x + 5y 3
s+r = + , (7)
10 5
with the relation
15y 2 − x2 = 14 , (8)
-135-
N = 134 N = 14
n xn yn xn yn xn yn xn yn
0 1 3 41 11 1 1 11 3
1 49 13 329 85 19 5 89 23
2 391 101 2591 669 151 39 701 181
3 3079 795 20399 5267 1189 307 5519 1425
4 24241 6259 160601 41467 9361 2417 43451 11219
5 190849 49277 1264409 326469 73699 19029 342089 88327
6 1502551 387957 9954671 2570285 580231 149815 2693261 695397
7 11829559 3054379 78372959 20235811 4568149 1179491 21203999 5474849
Table 2. First five solutions of recurrence relation (5) for which µ has integer values.
n xn yn µ
1 329 85 11
3 20399 5267 680
5 1264409 236469 42147
7 78372959 20235811 2612432
9 4857859049 1254293813 161928635
Since s + r = 3µ + y−3
there are at least + 1 non-isomorphic graphs having
6µ+y−3
2
, 4
Table 3. First five solutions of recurrence relation (9) for which µ has integer values.
n xn yn µ
1 83 233 3
3 5519 1425 184
5 342089 88327 11403
7 21203999 43103395 706800
9 1314305849 339352311 43810195
Since s + r = 3µ + y+1
there are at least + 1 non-isomorphic graphs having
6µ+y+1
2
, 4
and is one of the oldest and most thoroughly studied topological indices. It was introduced
in 1972 by Gutman and Trinajstić [14], surveyed 30 years after in [8], while 45 years after
-137-
the bounds for the Zagreb indices were surveyed in [3]. The first Zagreb index keeps going
to be in the focus of researchers, cf. [1, 2, 26]. The connection reads as follows.
Proof. Clearly, n(L(G)) = m(G), while the number of edges of L(G) can be expressed as
follows:
X deg(v) 1 X 1 X
m(L(G)) = = deg(v)2 − deg(v)
2 2 2
v∈V (G) v∈V (G) v∈V (G)
1
= M1 (G) − m(G) .
2
Since diam(L(G)) = 2, the following relation follows from [24, Theorem 2]:
n(L(G)) p(L(G))
P W (L(G)) = + − m(L(G)) .
2 2
Consequently,
m(G) p(L(G)) 1
P W (L(G)) = + − M1 (G) − m(G)
2 2 2
m(G) + 1 p(L(G)) 1
= + − M1 (G) .
2 2 2
For regular graphs Theorem 4.1 simplifies as follows. The computation is straightfor-
ward and hence omitted.
That is, in WP (G) only the distance between those pairs of vertices are counted which lie
in the same part of the partition P.
Now, let G be a graph with the periphery P (G) and let V (G) − P (G) = {v1 , . . . , vk }.
Let us call the partition T = {P (G), {v1 }, . . . , {vk }} of V (G) the peripheral partition.
Thus, if T is the peripheral partition of V (G), then we have:
P W (G) = WT (G) .
One of the main theorems from [18] then, for the case of the peripheral Wiener index,
reduces to the following result.
Theorem 5.1. Let G be a connected graph and {F1 , . . . , Fr } the Θ∗ -partition of E(G).
Then
r
X
P W (G) = W (G/Fi , wi ) ,
i=1
We do not formally define here the concepts used in Theorem 5.1, the reader can find
them in [18] or in [17], where a survey on the cut method is given. The method itself has
a long history, see [31] (references therein) for a recent nice development.
Let us conclude with a remark that the following result from [24, Theorem 5] is a very
special case of Theorem 5.1. If T is a tree and e ∈ E(T ), then let p1 (e) and p2 (e) be the
number of peripheral vertices of T lying on the two sides of e, respectively. Then:
Proposition 5.2. If T is a tree, then
X
P W (T ) = p1 (e) · p2 (e) .
e∈E(T )
6 Concluding remarks
In this paper we were in particular interested in graphs G for which P W (G) = P W (L(G))
holds. A more general (and also interesting) question is to find, or, ideally, characterize
-139-
References
[1] M. An, K. C. Das, First Zagreb index, k-connectivity, β-deficiency and k-
hamiltonicity of graphs, MATCH Commun. Math. Comput. Chem. 80 (2018)
141–151.
[2] L. Bedratyuk, O. Savenko, The star sequence and the general first Zagreb index,
MATCH Commun. Math. Comput. Chem. 79 (2018) 407–414.
[5] Y. H. Chen, H. Wang, X. D. Zhang, Peripheral Wiener index of trees and related
questions, Discr. Appl. Math. 251 (2018) 135–145.
[6] A. A. Dobrynin, The Szeged and Wiener indices of line graphs, MATCH Commun.
Math. Comput. Chem. 79 (2018) 743–756.
[7] E. Estrada, Generalized spectral moments of the iterated line graphs sequence. A
novel approach to QSPR studies, J. Chem. Inf. Comput. Sci. 39 (1999) 90–95.
-140-
[8] I. Gutman, K. C. Das, The first Zagreb index 30 years after, MATCH Commun.
Math. Comput. Chem. 50 (2004) 83–92.
[9] I. Gutman, V. Jovašević, A. A. Dobrynin, Smallest graphs for which the distance of
the graph is equal to the distance of its line graph, Graph Theory Notes New York
33 (1997) 19–19.
[10] I. Gutman, E. Estrada, Topological indices based on the line graph of the molecular
graph, J. Chem. Inf. Comput. Sci. 36 (1996) 541–543.
[11] I. Gutman, B. Furtula, A survey on terminal Wiener index, in: I. Gutman, B. Fur-
tula (Eds.), Novel Molecular Structure Descriptors-Theory and Applications I , Univ.
Kragujevac, Kragujevac, 2010, pp. 173–190.
[13] I. Gutman, Ž. Tomović, B. K. Mishra, M. Kuanar, On the use of iterated line graphs
in quantitative structure-property studies, Indian J. Chem. 40A (2001) 4–11.
[14] I. Gutman, N. Trinajstić, Graph theory and molecular orbitals. Total π-electron
energy of alternant hydrocarbons, Chem. Phys. Lett. 17 (1972) 535–538.
[15] H. Hua, On the peripheral Wiener index of graphs, Discr. Appl. Math., in press.
[21] D. J. Klein, I. Lukovits, I. Gutman, On the definition of the hyper–Wiener index for
cycle–containing structures, J. Chem. Inf. Comput. Sci. 35 (1995) 50–52.
-141-
[23] M. Knor, R. Škrekovski, A. Tepeh, Mathematical aspects of Wiener index, Ars Math.
Contemp. 11 (2016) 327–352.
[28] K. H. Rosen, Elementary Number Theory and its Applications, Fourth Edition,
Addison–Wesley, Reading, 2000.
[29] D. Shubhalakshmi, Distance Parameters and Its Applications, Ph.D. thesis, Manga-
lore University, Mangalagangothri, India 2017.
[31] N. Tratnik, A method for computing the edge-hyper-Wiener index of partial cubes
and an algorithm for benzenoid systems, Appl. Anal. Discr. Math. 12 (2018) 126–142.
[32] H. Wiener, Structural determination of paraffin boiling points, J. Am. Chem. Soc.
69 (1947) 17–20.
[33] Z. Yarahmadi, S. Moradi, The center and periphery of composite graphs, Iranian J.
Math. Chem. 5 (Supp. 1) (2014) s35–s44.