ISBN 978-1-59973-517-7

VOLUME 1, 2017

MATHEMATICAL COMBINATORICS
(INTERNATIONAL BOOK SERIES)

Edited By Linfan MAO

THE MADIS OF CHINESE ACADEMY OF SCIENCES AND

ACADEMY OF MATHEMATICAL COMBINATORICS & APPLICATIONS, USA

March, 2017
Vol.1, 2017 ISBN 978-1-59973-517-7

MATHEMATICAL COMBINATORICS
(INTERNATIONAL BOOK SERIES)

Edited By Linfan MAO

(www.mathcombin.com)

The Madis of Chinese Academy of Sciences and

Academy of Mathematical Combinatorics & Applications, USA

March, 2017
Aims and Scope: The Mathematical Combinatorics (International Book Series) is
a fully refereed international book series with ISBN number on each issue, sponsored by the
MADIS of Chinese Academy of Sciences and published in USA quarterly comprising 110-160
pages approx. per volume, which publishes original research papers and survey articles in all
aspects of Smarandache multi-spaces, Smarandache geometries, mathematical combinatorics,
non-euclidean geometry and topology and their applications to other sciences. Topics in detail
to be covered are:
Smarandache multi-spaces with applications to other sciences, such as those of algebraic
multi-systems, multi-metric spaces, , etc.. Smarandache geometries;
Topological graphs; Algebraic graphs; Random graphs; Combinatorial maps; Graph and
map enumeration; Combinatorial designs; Combinatorial enumeration;
Differential Geometry; Geometry on manifolds; Low Dimensional Topology; Differential
Topology; Topology of Manifolds; Geometrical aspects of Mathematical Physics and Relations
with Manifold Topology;
Applications of Smarandache multi-spaces to theoretical physics; Applications of Combi-
natorics to mathematics and theoretical physics; Mathematical theory on gravitational fields;
Mathematical theory on parallel universes; Other applications of Smarandache multi-space and
combinatorics.
Generally, papers on mathematics with its applications not including in above topics are
also welcome.
It is also available from the below international databases:
Serials Group/Editorial Department of EBSCO Publishing
10 Estes St. Ipswich, MA 01938-2106, USA
Tel.: (978) 356-6500, Ext. 2262 Fax: (978) 356-9371
http://www.ebsco.com/home/printsubs/priceproj.asp
and
Gale Directory of Publications and Broadcast Media, Gale, a part of Cengage Learning
27500 Drake Rd. Farmington Hills, MI 48331-3535, USA
Tel.: (248) 699-4253, ext. 1326; 1-800-347-GALE Fax: (248) 699-8075
http://www.gale.com
Indexing and Reviews: Mathematical Reviews (USA), Zentralblatt Math (Germany), Refer-
ativnyi Zhurnal (Russia), Mathematika (Russia), Directory of Open Access (DoAJ), Interna-
tional Statistical Institute (ISI), International Scientific Indexing (ISI, impact factor 1.730),
Institute for Scientific Information (PA, USA), Library of Congress Subject Headings (USA).
Subscription A subscription can be ordered by an email directly to
Linfan Mao
The Editor-in-Chief of International Journal of Mathematical Combinatorics
Chinese Academy of Mathematics and System Science
Beijing, 100190, P.R.China
Email: [email protected]

Price: US$48.00
Editorial Board (4th)

Editor-in-Chief

Linfan MAO Shaofei Du

Chinese Academy of Mathematics and System Capital Normal University, P.R.China
Science, P.R.China Email: [email protected]
and
Xiaodong Hu
Academy of Mathematical Combinatorics &
Chinese Academy of Mathematics and System
Applications, USA
Science, P.R.China
Email: [email protected]
Email: [email protected]
Deputy Editor-in-Chief Yuanqiu Huang
Hunan Normal University, P.R.China
Guohua Song Email: [email protected]
Beijing University of Civil Engineering and
H.Iseri
Architecture, P.R.China
Mansfield University, USA
Email: [email protected]
Email: [email protected]

Editors Xueliang Li
Nankai University, P.R.China
Arindam Bhattacharyya Email: [email protected]
Jadavpur University, India Guodong Liu
Email: [email protected] Huizhou University
Said Broumi Email: [email protected]
Hassan II University Mohammedia W.B.Vasantha Kandasamy
Hay El Baraka Ben Msik Casablanca Indian Institute of Technology, India
B.P.7951 Morocco Email: [email protected]
Junliang Cai Ion Patrascu
Beijing Normal University, P.R.China Fratii Buzesti National College
Email: [email protected] Craiova Romania
Yanxun Chang Han Ren
Beijing Jiaotong University, P.R.China East China Normal University, P.R.China
Email: [email protected] Email: [email protected]
Jingan Cui Ovidiu-Ilie Sandru
Beijing University of Civil Engineering and Politechnica University of Bucharest
Architecture, P.R.China Romania
Email: [email protected]
ii Mathematical Combinatorics (International Book Series)

Mingyao Xu
Peking University, P.R.China
Email: [email protected]

Guiying Yan
Chinese Academy of Mathematics and System
Science, P.R.China
Email: [email protected]
Y. Zhang
Department of Computer Science
Georgia State University, Atlanta, USA

Famous Words:

You can pay attention to the fact, in which case youll probably become a math-
ematician, or you can ignore it, in which case youll probably become a physicist.

By Len Evans, an American mathematician.

Math.Combin.Book Ser. Vol.1(2017), 01-09

Special Smarandache Curves According to Bishop

Frame in Euclidean Spacetime

E. M. Solouma
(Department of Mathematics, Faculty of Science, Beni-Suef University, Egypt)

M. M. Wageeda
(Mathematics Department, Faculty of Science, Aswan University, Aswan, Egypt)

E-mail: [email protected], [email protected]

Abstract: In this paper, we introduce some special Smarandache curves according to

Bishop frame in Euclidean 3-space E3 . Also, we study Frenet-Serret invariants of a special
case in E3 . Finally, we give an example to illustrate these curves.

Key Words: Smarandache curve, Bishop frame, Euclidean spacetime.

AMS(2010): 53A04, 53A05.

1. Introduction

In the theory of curves in the Euclidean and Minkowski spaces, one of the interesting problems
is the characterization of a regular curve. In the solution of the problem, the curvature functions
and of a regular curve have an effective role. It is known that the shape and size of a regular
curve can be determined by using its curvatures and ([7],[8]). For instance, Bertrand curves
and Mannheim curves arise from this relationship. Another example is the Smarandache curves.
They are the objects of Smarandache geometry, that is, a geometry which has at least one
Smarandachely denied axiom [1]. The axiom is said to be Smarandachely denied if it behaves
in at least two different ways within the same space. Smarandache geometries are connected
with the theory of relativity and the parallel universes.
By definition, if the position vector of a curve is composed by the Frenet frames vectors
of another curve , then the curve is called a Smarandache curve [9]. Special Smarandache
curves in the Euclidean and Minkowski spaces are studied by some authors ([6], [10]). For
instance, the special Smarandache curves according to Darboux frame in E3 are characterized
in [5].
In this work, we study special Smarandache curves according to Bishop frame in the Eu-
clidean 3-space E3 . We hope these results will be helpful to mathematicians who are specialized
on mathematical modeling.

1 Received August 23, 2016, Accepted February 2, 2017.

2 E. M. Solouma and M. M. Wageeda

2. Preliminaries

The Euclidean 3-space E3 provided with the standard flat metric given by

h , i = dx21 + dx22 + dx23 ,

where (x1 , x2 , x3 ) is a rectangular coordinate system of E3 . Recall that, the norm of an arbitrary
vector v E3 is given by kvk = |hv, vi|. A curve is called an unit speed curve if velocity
p

vector of satisfies k k = 1. For vectors u, v E3 it is said to be orthogonal if and only if

hu, vi = 0. Let = (s) be a regular curve in E3 . If the tangent vector field of this curve forms
a constant angle with a constant vector field U , then this curve is called a general helix or an
inclined curve.

Denote by {T, N, B} the moving Frenet frame along the curve in the space E3 . For
an arbitrary curve E3 , with first and second curvature, and respectively, the Frenet
formulas is given by ([7]).

T (s) 0 0 T (s)

N (s) = 0
N (s) ,
(1)


B (s) 0 0 B(s)

where hT, T i = hN, N i = hB, Bi = 1, hT, N i = hT, Bi = hN, Bi = 0. Then, we write Frenet
invariants in this way: T (s) = (s), (s) = kT (s)k, N (s) = T (s)/(s), B(s) = T (s) N (s)
and (s) = hN (s), B (s)i.

The Bishop frame or parallel transport frame is an alternative approach to defining a

moving frame that is well defined even when the curve has vanishing second derivative. One
can express Bishop of an orthonormal frame along a curve simply by parallel transporting each
component of the frame [2]. The tangent vector and any convenient arbitrary basis for the
remainder of the frame are used (for details, see [3]). The Bishop frame is expressed as ([2],
[4]).

T (s) 0 k1 (s) k2 (s) T (s)

N (s) = k1 (s) 0 0 N1 (s) . (2)

1
N2 (s) k2 (s) 0 0 N2 (s)

Here, we shall call the set {T, N1 , N2 } as Bishop trihedra and k1 (s) and k2 (s) as Bishop curva-
tures. The relation matrix may be expressed as

T (s) 1 0 0 T (s)

N1 (s) = 0 cos (s)
N (s) ,
sin (s) (3)


N2 (s) 0 sin (s) cos (s) B(s)
 Special Smarandache Curves According to Bishop Frame in Euclidean Spacetime 3

where k 
2

(s) = arc tan , k1 6= 0
k1



d(s) (4)
(s) =


ds
(s) = pk 2 (s) + k 2 (s)

1 2

Here, Bishop curvatures are defined by

k (s) = (s) cos (s),
1
(5)
k2 (s) = (s) sin (s).

Let = (s) be a regular non-null curve parametrized by arc-length in Euclidean 3-space

E3 with its Bishop frame {T, N1 , N2 }. Then T N1 , T N2 , N1 N2 and T N1 N2 -Smarandache curve
of are defined, respectively as follows ([9]):

1  
B = B((s)) = T (s) + N1 (s) ,
2
1  
B = B((s)) = T (s) + N2 (s) ,
2
1  
B = B((s)) = N1 (s) + N2 (s) ,
2
1  
B = B((s)) = T (s) + N1 (s) + N2 (s) .
3

3. Special Smarandache Curves According to Bishop Frame in E3

Definition 3.1 A regular curve in Euclidean space-time, whose position vector is composed by
Frenet frame vectors on another regular curve, is called a Smarandache curve.

In the light of the above definition, we adapt it to regular curves according to Bishop frame
in the Euclidean 3-space E3 as follows.

Definition 3.2 Let = (s) be a unit speed regular curve in E3 and {T, N1 , N2 } be its moving
Bishop frame. T N1 -Smarandache curves are defined by

1
B = B((s)) = (T (s) + N1 (s)) . (6)
2

Let us investigate Frenet invariants of T N1 -Smarandache curves according to = (s).

By differentiating Eqn.(6) with respected to s and using Eqn.(2), we get

dB d 1  
B = = k1 T + k1 N1 + k2 N2 , (7)
d ds 2
4 E. M. Solouma and M. M. Wageeda

and hence
k1 T + k1 N1 + k2 N2
TB = p , (8)
2k12 + k22
where r
d 2k12 + k22
= . (9)
ds 2
In order to determine the first curvature and the principal normal of the curve B, we
formalize
dTB d d 1 T + 2 N1 + 3 N2
TB = = TB =
3 , (10)
d ds ds 2k12 + k22 2
where
2 2 2
k22 ) ,
 
1 = k1 (2k1 k1 + k2 k2 ) (2k1 + k2 )(k1 + k1 +


2 = (2k12 + k22 )(k1 k12 ) k1 (2k1 k1 + k2 k2 ) ,

(11)


3 = (2k12 + k22 )(k2 k1 k2 ) k2 (2k1 k1 + k2 k2 ) .
 

Then, we have
2  
TB = 1 T + 2 N 1 + 3 N 2 . (12)
2k12 + k2 2 2

So, the first curvature and the principal normal vector field are respectively given by
p 2
2 1 + 22 + 32
B = kTB k =
2 . (13)
2k12 + k22

and
1 T + 2 N1 + 3 N2
NB = p . (14)
12 + 22 + 32
On other hand, we express
 
 T N1 N2 

1  
T B NB = k1 k1 k2  , (15)
p q  
 1 2 3 
p p
where p = 2k12 + k22 and q = 12 + 22 + 32 . So, the binormal vector is

1 n      o
BB = k1 3 k2 2 T + k1 3 + k2 1 N1 + k1 1 + 2 N2 . (16)
pq

In order to calculate the torsion of the curve B, we differentiate Eqn.(7) with respected to
s, we have

1 n   o
B = k1 + k12 + k1 k2 + T + k1 k12 1 N1 + k2 k1 k2 N2 .
   
(17)
2

and thus
1 T + 2 N1 + 2 N2
B = , (18)
2
 Special Smarandache Curves According to Bishop Frame in Euclidean Spacetime 5

where
= k1 + k1 (3k1 + k2 ) + k2 (k1 + k2 ) k1 (k12 + k22 ) ,
 
1


2 = k1 k1 (k12 + 3k1 + k1 k2 ), (19)


3 = k2 k1 k2 k2 (k12 + 2k1 + k1 k2 ).

The torsion is then given by:

 2
2 (k1 k1 )(k1 3 + k2 1 ) + k1 (k2 k1 k2 )(1 + 2 ) + (k12 + k1 + k1 k2 )(k1 3 k2 2 )

B =  2
. (20)
(k1 k2 k1 k2 )2 + k1 k2 + k2 (k1 + k1 k2 ) + k12 (2k12 + k1 k2 )2

Corollary 3.1 Let = (s) be a curve lying fully in E3 with the moving frame {T, N, B}. If
is contained in a plane, then the Bishop curvatures becomes constant and the T N1 -Smarandache
curve is also contained in a plane and its curvature satisfying the following equation
q 
2 k12 (k22 + 1) + (k12 + k22 )2

B = .
2k12 + k22
.
Definition 3.3 Let = (s) be a unit speed regular curve in E3 and {T, N1 , N2 } be its moving
Bishop frame. T N2 -Smarandache curves are defined by

1  
B = B((s)) = T (s) + N2 (s) . (21)
2

Remark 3.1 The Frenet invariants of T N2 -Smarandache curves can be easily obtained by the
apparatus of the regular curve = (s).

Definition 3.4 Let = (s) be a unit speed regular curve in E3 and {T, N1 , N2 } be its moving
Bishop frame. N1 N2 -Smarandache curves are defined by

1  
B = B((s)) = N1 (s) + N2 (s) . (22)
2

Remark 3.2 The Frenet invariants of N1 N2 -Smarandache curves can be easily obtained by
the apparatus of the regular curve = (s).

Definition 3.5 Let = (s) be a unit speed regular curve in E3 and {T, N1 , N2 } be its moving
Bishop frame. T N1 N2 -Smarandache curves are defined by

1  
B = B((s)) = T (s) + N1 (s) + N2 (s) . (23)
3

Remark 3.3 The Frenet invariants of T N1 N2 -Smarandache curves can be easily obtained by
the apparatus of the regular curve = (s).
Example 3.1 Let (s) = 12 cos s, sin s, s be a curve parametrized by arc length. Then

it is easy to show that T (s) = 12 sin s, cos s, 1 , = 12 6= 0, tanh = 12 6= 0 and (s) =

1 s + c, c = constant. Here, we can take c = 0. From Eqn.(4), we get k1 (s) = 1 cos s ,

2 2 2
6 E. M. Solouma and M. M. Wageeda

Z Z
1 s


k2 (s) = 2
sin 2
. From Eqn.(1), we get N1 (s) = k1 (s)T (s)ds, N2 (s) = k2 (s)T (s)ds,
then we have

 

2  2 
N1 (s) = cos (1 + 2)s cos (1 2)s ,
4(1 + 2) 4(1 2)

2   2   2  s 
sin (1 + 2)s sin (1 2)s , sin
4(1 + 2) 4(1 2) 2 2



2   2  
N2 (s) = sin (1 + 2)s sin (1 2)s ,
4(1 + 2) 4(1 2)

2   2   2  s 
cos (1 + 2)s + cos (1 2)s , cos .
4(1 + 2) 4(1 2) 2 2

0.5

0.0

-0.5

-0.5

0.0

0.5

Figure 1 The curve = (s).

In terms of definitions, we obtain special Smarandache curves, see Figures 2 - 5.

1.0

0.5

0.0

-0.5

1.0

0.5

0.0
0.0

0.5

1.0

Figure 2 T N1 -Smarandache curve.

 Special Smarandache Curves According to Bishop Frame in Euclidean Spacetime 7

0.0

-0.5

-1.0

1.0

0.5

0.0
0.0

0.5

1.0

Figure 3 T N2 -Smarandache curve.

-0.6 0.8
1.0
-0.8 1.2
1.4
-1.0

-1.2

0.5

0.0

-0.5

Figure 4 N1 N2 -Smarandache curve.

8 E. M. Solouma and M. M. Wageeda

0.0
-0.2

-0.4 0.5
-0.6
1.0
-0.8

1.0

0.5

0.0

Figure 5 T N1 N2 -Smarandache curve.

4. Conclusion

Consider a curve = (s) parametrized by arc-length in Euclidean 3-space E3 that the curve
(s) is sufficiently smooth so that the Bishop frame adapted to it is defined. In this paper, we
study the problem of constructing Frenet-Serret invariants {TB , NB , BB , B , B } from a given
some special curve B according to Bishop frame in Euclidean 3-space E3 that posses this curve
as Smarandache curve. We list an example to illustrate the discussed curves. Finally, we hope
these results will be helpful to mathematicians who are specialized on mathematical modeling.

References

[1] C. Ashbacher, Smarandache geometries, Smarandache Notions Journal, Vol. 8 (13) (1997)
, 212215.
[2] L.R. Bishop, There is more than one way to frame a curve, Amer. Math. Monthly, 82 (3)
(1975), 246251.
[3] B. Bukcu, M. K. Karacan, Parallel transport frame of the spacelike curve with a spacelike
binormal in Minkowski 3-space, Selcuk J. Appl. Math., 11 (1) (2010), 1525.
[4] B. Bukcu, M. K. Karacan, Bishop frame of the spacelike curve with a spacelike binormal
in Minkowski 3-space, Selcuk J. Appl. Math., 11 (1) (2010), 1525.
Bektas, S. Yce, Special Smarandache curves according to Darboux frame in Euclidean
[5] O.
3- Space, Romanian Journal of Mathematics and Computer Science, 3 (2013), 4859.
[6] M. C
etin, Y. Tuncer, M. K. Karacan, Smarandache curves according to Bishop frame in
Euclidean 3- space, General Mathematics Notes, 20 (2) (2014), 5066.
[7] M.P. Do Carmo, Differential Geometry of Curves and Surfaces, Prentice Hall, Englewood
 Special Smarandache Curves According to Bishop Frame in Euclidean Spacetime 9

Cliffs, NJ, 1976.

[8] B. O.Neill, Elementary Differential Geometry, Academic Press Inc. New York, 1966.
[9] M. Turgut, S. Ylmaz, Smarandache curves in Minkowski space-time, International Journal
of Mathematical Combinatorics, 3 (2008), 5155.
[10] K. Taskopr
u and M. Tosun, Smarandache curves according to Sabban frame on, Boletim
da Sociedade Paraneanse de Matematica, 32 (1) (2014), 5159.
Math.Combin.Book Ser. Vol.1(2017), 10-21

Spectra and Energy of Signed Graphs

Nutan G. Nayak
(Department of Mathematics and Statistics, S. S. Dempo College of Commerce and Economics, Goa, India)

E-mail: [email protected]

Pn
Abstract: The energy of a signed graph is defined as () = i=1 |i |, where
1 , 2 , , n are the eigenvalues of . In this paper, we study the spectra and energy
of a class of signed graphs which satisfy pairing property. We show that it is possible to
compare the energies of a pair of bipartite and non-bipartite signed graphs on n vertices
by defining quasi-order relation in such a way that the energy is increasing. Further, we
extend the notion of extended double cover of graphs to signed graphs to find the spectra
of unbalanced signed bipartite graphs and also we construct non-cospectral equienergetic
signed bipartite graphs.

Key Words: Signed graph, Smarandachely signed graph, signed energy, extended double
cover(EDC) of signed graphs, equienergetic signed bipartite graphs.

AMS(2010): 05C22, 05C50.

1. Introduction

A signed graph is an ordered pair = (G, ), where G is the underlying graph of and
: E {+1, 1}, called signing (or a signature), is a function from the edge set E(G) of
G into the set {+1, 1}. It is said to be homogeneous if its edges are all positive or negative
otherwise heterogeneous, and a Smarandachely signed if |e+ e | 1, where e+ , e are
numbers of edges signed by +1 or 1 in E(G), respectively. Negation of a signed graph is the
same graph with all signs reversed. In figure, we denote positive edges with solid lines and
negative edges with dotted lines.
The adjacency matrix of a signed graph is the square matrix A() = (aij ) where (i, j) entry
is +1 if (vi vj ) = +1 and 1 if (vi vj ) = 1, 0 otherwise. The characteristic polynomial of the
signed graph is defined as ( : ) = det(I A()), where I is an identity matrix of order
n. The roots of the characteristic equation ( : ) = 0, denoted by 1 , 2 , , n are called
the eigenvalues of signed graph . If the distinct eigenvalues of A() are 1 > 2 > > n
and their multiplicities are m1 , m2 , , mn then the spectrum of is

1 2 ... ... n
Spec() = .
m1 m2 ... ... mn

1 Received August 16, 2016, Accepted February 3, 2017.

 Spectra and Energy of Signed Graphs 11

Two signed graphs are cospectral if they have the same spectrum. The spectral criterion
for balance in signed graph is given by B. D. Acharya as follows:

Theorem 1.1([1]) A signed graph is balanced if and only if it is cospectral with the underlying
graph. i.e. Spec() = Spec(G).

The sign of a cycle in a signed graph is the product of the signs of its edges. Thus a cycle
is positive if and only if it contains an even number of negative edges. A signed graph is said
to be balanced if all of its cycles are positive otherwise unbalanced.
In a signed graph , the degree of a vertex v is defined as sdeg(v) = d(v) = d+
(v) + d (v),
+
where d (v)(d (v)) is the number of positive(negative) edges incident with v. It is said to be
regular if all its vertices have same degree. The net degree of a vertex v of a signed graph
is d +
(v) = d (v) d (v). It is said to be net-regular of degree k if all its vertices have same
net-degree equal to k.
Spectra of graphs is well documented in [5] and signed graphs is discussed in [7, 8, 9, 11].
For standard terminology and notations in graph theory we follow D. B. West [15] and for
signed graphs we follow T. Zaslavsky [16].
If 1 , 2 , , n are the eigenvalues of , then () = ni=1 |i |. Two signed graphs 1
P

and 2 are said to be equienergetic if (1 ) = (2 ). Naturally, cospectral signed graphs are

equienergetic. Equienergetic signed graphs are constructed in [3, 13].
The cartesian product 1 2 of two signed graphs 1 = (V1 , E1 , 1 ) and 2 = (V2 , E2 , 2 )
is defined as the signed graph (V1 V2 , E, ) where the edge set E is that of the Cartesian product
of the underlying unsigned graphs and the signature function for the labeling of the edges is
defined by
(u , u ), if j = l
1 i k
((ui , vj )(uk , vl )) =
2 (vj , vl ), if i = k
N
The Kronecker product of 1 2 of two signed graphs 1 = (V1 , E1 , 1 ) and 2 =
(V2 , E2 , 2 ) is the signed graph (V1 V2 , E, ) where the edge set E is that of the Kronecker
product of the underlying unsigned graphs and the signature function for the labeling of the
edges is defined by ((ui , vj )(uk , vl )) = 1 (ui , uk )2 (vj , vl ).
Generally, quasi-order relation is used to compare the energies of bipartite graphs. In this
paper, we use quasi-order method to compare the energies of two signed graphs of order n which
are bipartite and unbalanced non-bipartite signed graphs. Fundamental question in the energy
theory is to find the maximal and minimal energy graphs over a significant class of graphs. It
is natural to find for signed graphs also. Here we give maximum energy signed graphs which
belong to the class of pairing property. Further, we study the spectra and energy of extended
double cover (EDC) of signed graphs and also construct non-cospectral equienergetic signed
bipartite graphs.

2. Energy of Signed Graphs in n

A graph G is a bipartite graph if and only if i = n+1i , for 1 i 12 (n 1). This result
12 Nutan G. Nayak

is known as pairing theorem by Coulson and Rushbrooke [6]. But non-bipartite signed graphs
also satisfy pairing property and examples are given in [3]. The class of signed graphs satisfying
pairing property we denote it as n .
The following result is given by Bhat and Pirzada in [3] which gives the spectral criterion
of signed graphs on n .

Theorem 2.1 Let be a signed graph of order n which satisfies the pairing property. Then
the following statements are equivalent:

(1) spectrum of is symmetric about the origin;

P
2
n

n
(2) () = + (1)k b2k n2k , where b2k are non-negative integers for all k =
k=1
1, 2, , n2 ;
 

(3) and are cospectral, where is the signed graph obtained by negating sign of
each edge of .

Now it is possible to define a quasi-order relation over n in such a way that the energy is
increasing. Note that n consists of signed bipartite as well as unbalanced non-bipartite signed
graphs which satisfy pairing property.

Definition 2.2 Let 1 and 2 be two signed graphs of order n in n . From Theorem 2.1 we
can express
X
2
n

n
() = + (1)k b2k n2k
k=1

where b2k are non-negative integers for all k = 1, 2, , n2 . If b2k (1 ) b2k (2 ) for all k
 

where 1 k n2 then we can write 1 2 . Further, if b2k (1 ) < b2k (2 ) for all k where
 
n
1 k 2 then we write 1 < 2 . Hence

1  2 (1 ) (2 ),

1 2 (1 ) < (2 ),

which implies that the energy is increasing in a quasi order relation over n .

In [13], it is shown that Coulsons Integral formula remains valid for signed graphs also.

Theorem 2.3([13]) If is a signed graph then the energy of signed graph is

" #
+
1 i (i)
Z
() = n d.
(i)

Following result is the consequence of Coulsons Integral formula for signed graphs.
 Spectra and Energy of Signed Graphs 13

Corollary 2.4 Let be a signed graph. Then

+  
1 1 i
Z
k
() = ln ( ) d.
2

Theorem 2.5 Let n . Then the energy of a signed graph can be expressed as

+ X
2
n

1 1
Z
() = ln 1 + (1)k b2k ()2k d.

2
k=1

and if 1 , 2 n and 1 < 2 then (1 ) < (2 ).

Proof Coulsons Integral formula can be expressed as

+  
1 1 i
Z
k
() = ln ( ) d.
2

Since n , from Theorem 2.1 we can deduce

   n X
2
n

i i
= 1 + (1)k b2k ()2k

n
k=1

and substituting in the above expression, we get

+ 2
X
n

1 1 n
Z
() = ln i (1 + (1)k b2k ()2k ) d

2
k=1

+ X
2
n

1 1
Z
= ln 1 + (1)k b2k ()2k d.

2
k=1

R + 1
But 1 p.v. 2 ln[in ]d = 0 where p.v. is the principal value of Cauchys integral.

Hence () is a monotonically increasing function on the coefficients of b2k (). 2
Now the question is which signed graphs are having maximum signed energy in n .

Theorem 2.6([14]) Let be a signed graph with n vertices and m edges, then
q
2m + n(n 1)| det(A())|2/n () 2mn.

Corollary 2.7 () = 2mn = n r if and only if T = ()2 = rIn , where r is the maximum
degree of and In is the identity matrix of order n.

Proof Notice that () = n r if and only if there exists a constant t such that |i |2 = t for
14 Nutan G. Nayak

all i and is an r-regular signed graph. Hence equality holds if and only if T = ()2 = tI
and t = r. 2
The following two examples are given by the present author in [12, 14].

Example 2.8 Following unbalanced signed cycle, we denote it as (C4 ).

v1 v2

v4 v3

Fig.1 Signed cycle with maximum signed energy

0 1 0 1

1 0 1 0
A(C4 ) =


0 1 0 1

1 0 1 0

The characteristic polynomial is (C4 ) = 4 42 + 4 and Spec (C4 )={( 2)2 , ( 2)2 }

n . Hence (C4 ) = 4 2 = n r.

Example 2.9 Following unbalanced signed complete graph, we denote it as (K6 ).

v1

v6 v2

v5 v3

v4

Fig.2 Signed Complete graph with maximum signed energy

0 1 1 1 1 1


1 0 1 1
1
1

1 1 0 1 1 1
A(K6 ) =



1 1 1 0 1 1


1 1 1 1 0 1
1 1 1 1 1 0
 Spectra and Energy of Signed Graphs 15

which is a symmetric conference matrix having the characteristic polynomial (K6 ) = 5

153 + 75 125 and Spec A(K6 )={( 5)3 , ( 5)3 } n . The signed energy of (K6 ) =

6 5 = n r.

Lemma 2.10([3, 8]) Let 1 and 2 be two signed graphs with eigenvalues 1 , 2 , , n1 and
1 , 2 , , n2 . Then

(1) the eigenvalues of 1 2 = i + j , for all i = 1, 2, , n1 , j = 1, 2, , n2 ;

N
(2) the eigenvalues of 1 2 = i j , for all i = 1, 2, , n1 , j = 1, 2, , n2 .

Theorem 2.11 There exists an infinite family of signed graphs having maximum signed energy
in n .

Proof Let 1 , 2 be two signed graphs in n with orders n1 and n2 having maximum
N
energies n1 r1 , n2 r2 respectively. The Kronecker product of 1 2 is a symmetric matrix

2
N
of order n1 n2 . From Lemma 2.10, 1 2 has maximum energy n1 n2 r1 r2 .

Here we note that maximum energy signed graphs belong to the class of n .

3. Spectra of Signed Biparitite Graphs in n .

In [2], N. Alon introduced the concept of extended double cover of a graph. Here we extend this
notion to signed graphs in order to establish the spectrum of various signed bipartite graphs.
The ordinary spectrum of EDC of graph is given by Z. Chen in [4].

Lemma 3.1([4]) Let 1 , 2 , , n be the eignenvalues of the graph G. Then the eigenvalues
of extended double cover of graph are (1 + 1), (2 + 1), , (n + 1).

Now we define extended double cover of signed graph as follows:

Definition 3.2 Let be a signed graph with vertex set {v1 , v2 , , vn }. Let be a signed
bipartite graph with V ( ) = {v1 , v2 , , vn , u1 , u2 , , un } where,

1) vi is adjacent to ui and either (vi ui ) = +1 or (vi ui ) = 1;

2) vi is adjacent to uj if vi is adjacent to vj in ;

3) (vi uj ) = +1 if (vi vj ) = +1 and (vi uj ) = 1 if (vi vj ) = 1.

16 Nutan G. Nayak

v1

v3 P v2

v1 v2 v3 v1 v2 v3

u1 u2 u3 u1 u2 u3
P P
1 2

Fig.3 Extended double covers of signed graph .

Then is known as extended double cover of signed graph of signed graph and in short we
write it as EDC of . Since we get two EDCs of signed graph, we denote it as 1 if (vi ui ) = +1
and 2 if (vi ui ) = 1.
We need the following Lemma from [10] for further investigation.

A0 A1
Lemma 3.3([10]) Let A = be a symmetric 2 2 block matrix. Then the
A1 A0
spectrum of A is the union of the spectra of A0 + A1 and A0 A1 .

The following Lemma gives the relation between the spectrum of a signed graph and its
EDC of signed graph.

Lemma 3.4 Let 1 , 2 , , n be the eigenvalues of a signed graph then the spectrum of EDCs
of signed graph is
n o
(1) Spec(1 ) = (1 + 1), (2 + 1), , (n + 1)
n o
(2) Spec(2 ) = (1 1), (2 1), , (n 1)

Proof Let the adjacencymatrix of the signed

graph
be A. Thenthe adjacency matrix of
0 A+I 0 AI
EDC of signed graph of is or , where I is an identity
A+I 0 AI 0
matrix.
From Lemma 3.3, it is clear that the eigenvalues of are (i + 1) if (vi ui ) = +1 and
(i 1) if (vi ui ) = 1 for each eigenvalue of . 2
Theorem 3.5 Let be a connected signed graph. Then 1 , 2 and ( K2 ) are co-spectral
if and only if belongs to the class of n .

Proof Let 1 , 2 , , n be the eigenvalues of a signed graph then

n o
(i) Spec(1 ) = (1 + 1), (2 + 1), , (n + 1) ;
n o
(ii) Spec(2 ) = (1 1), (2 1), , (n 1) ;
 Spectra and Energy of Signed Graphs 17

n o
(iii) Spec( K2 ) = (1 1), (2 1), ..... , (n 1) .

So, Spec(1 ) = Spec(2 ) = Spec( K2 ) if and only if i = n+1i , for i = 1, 2, , n.

Hence the proof. 2
Now we give spectra of various signed bipartite graphs.

Proposition 3.6 Let (Pn )1 and (Pn )2 be the extended double covers of signed path Pn . Then
the spectrum is

i
(2cos n+1 + 1)
(1) Spec(Pn )1 = , i = 1, , n.
n

i
(2cos n+1 1)
(2) Spec(Pn )2 = , i = 1, , n.
n

Remark 3.7 If is a signed path then EDCs of signed paths are balanced. Hence EDCs of
signed paths are having same energy as underlying graph.

Proposition 3.8 Let Cn+ (Cn ) be the positive(negative) signed cycles on Cn . Then the spectrum
of EDCs are respectively

(1) If n is odd, then Spec(Cn+ )1 = (2cos 2i +

 
n + 1), i = 1, 2, , n and Spec(Cn )2 =
2i
 
(2cos n 1), i = 1, 2, , n ;
h i
(2) If n is even, then (i)Spec(Cn )1 = (2cos (2i+1)
n + 1), i = 1, 2, , n and Spec(Cn )2 =
h i
(2cos (2i+1)
n 1), i = 1, 2, , n .

If the signed graph is +Kn then EDCs of +Kn are (Kn )1 = +Kn,n and (Kn )2 . Spec(Kn,n )
= {n, 02n2 }. Following result gives the spectrum of (Kn )2 which is an unbalanced net-regular
signed complete bipartite graph.

Proposition 3.9 Let (Kn )2 be the EDC of +Kn . Then the spectrum of (Kn )2 is

2 k k 2
Spec(Kn )2 = ,
n1 1 1 n1

where k = d (Kn )2 = n 2.

Remark 3.10 From above Proposition 3.9, (Kn )2 = 2(3n 8).

Theorem 3.11([13]) The spectrum of heterogeneous unbalanced signed complete graph (Knnet )
is
2i
5 n 1 + 4 cos( n )
Spec(Knnet ) = , i = 1, , n 1.
1 1

where (Knnet ) is a net regular signed complete graph defined on +Kn .

18 Nutan G. Nayak

Proposition 3.12 If (Knnet )1 and (Knnet )2 are the net-regular signed complete bipartite graphs
of EDCs of Knnet . Then the spectrum is

(1)
k (2 + 4 cos( 2i
n ))
Spec(Knnet )1 = , i = 1, , (n 1),
1 n1

where k = (6 n) gives net regularity of (Knnet )1 .

(2)
k (1 + 4 cos( 2i
n ))
Spec(Knnet )2 = , i = 1, , n 1,
1 n1

where k = (4 n) gives net regularity of (Knnet )2 .

From the above Propositions, we are having the following result.

Theorem 3.13 EDCs of signed graphs are net-regular if and only if signed graph is net-
regular.

4. Equienergetic Signed Graphs in n

Here we construct equienergetic signed bipartite graphs on 4n vertices which are non-cospectral
and equienergetic.

Theorem 4.1 There exists a pair of non-cospectral equienergetic signed bipartite graphs on 4n
vertices where n is odd and n 3.

Proof Let be a signed cycle of order n and of odd length with eigenvalues 1 , 2 , , n
and let the extended double covers of signed graph be 1 and 2 .

Case 1. If is balanced then

2 i
Spec() = , i = 1, , n 1.
1 n1

By Lemma 3.4,

3 (i + 1)
Spec(1 ) = , i = 1, , n 1.
1 n1

and
1 (i 1)
Spec(2 ) = , i = 1, , n 1.
1 n1

Hence 1 and 2 are non-cospectral bipartite signed graphs on 2n vertices where n is odd
 Spectra and Energy of Signed Graphs 19

and 1 is balanced and 2 is unbalanced.

Further, let H1 , H2 and K1 , K2 be second iterated extended double cover signed graphs of
1 and 2 respectively. By Theorem 3.5, Spec H1 = Spec H2 and Spec K1 = Spec K2 . Let
Spec S = Spec H1 = Spec H2 and Spec T= Spec K1 = Spec K2 .

(4) (2) ((i + 1) + 1)
Spec(S) = , i = 1, , n 1.
1 1 2(n 1)

and
(2) (0) ((i 1) + 1)
Spec(T ) = , i = 1, , n 1.
1 1 2(n 1)

Hence S = (1 ) and T = (2 ) are non-cospectral bipartite signed graphs on 4n vertices

where n is odd.
n1
X
(S) = 2[4 + 2 + | (i + 1) + 1|],
i=1

n1
X
(T ) = 2[2 + 0 + | (i 1) + 1|].
i=1
Pn1
If (S) = (T ) then 4 = i=1 (| (i 1) + 1 | | (i + 1) + 1 |), then we know that

n1
X
4= (|2 i | + |i | |i + 2| |i |),
i=1

n1
X
4= (|i 2| |i + 2|).
i=1

Since is a balanced signed cycle i = 2cos 2i

n , i = 1, , n 1,

n1
X
4= (|2cosi 2| |2cosi + 2|),
i=1

n1
X i i
1= (sin2 ( ) cos2 ( )),
i=1
2 2
n1
1X
1 = 2cosi .
2 i=1
Pn1
Since i=1 i = 2, so (S) = (T ).

Case 2. If is unbalanced then

2 i
Spec() = , i = 1, , n 1.
1 n1
20 Nutan G. Nayak

By Lemma 3.4,

1 (i + 1)
Spec(1 ) = , i = 1, , n 1.
1 n1

and
3 (i 1)
Spec(2 ) = , i = 1, , n 1.
1 n1

By a similar argument as in Case 1, we get (S) = (T ). Hence the proof. 2

Example 4.2 Consider the signed graphs 1 and 2 as shown in Fig.3. By Lemma 3.4, the
characteristic polynomials of 1 and 2 are

(1 ) = ( + 2)2 ( 2)2 ( + 1)( 1)

(2 ) = 4 ( + 3)( 3)

The characteristic polynomials of (1 ) and (2 ) are

(1 ) = 2 ( + 1)2 ( 1)2 ( + 3)2 ( 3)2 ( + 2)( 2),

(2 ) = ( + 1)4 ( 1)4 ( + 4)( 4)( + 2)( 2).

Hence Spec(1 ) 6= Spec(2 ) but (1 ) = (2 ) = 20.

Another example of equienergetic signed bipartite graphs on 4n vertices is given below.

Example 4.3 Consider the signed graphs 1 and 2 as shown in Fig.3. By Lemma 2.10, the
characteristic polynomials of (1 K2 ) and (2 K2 ) are

(1 K2 ) = 2 ( + 1)2 ( 1)2 ( + 3)2 ( 3)2 ( + 2)( 2),

(2 K2 ) = ( + 1)4 ( 1)4 ( + 4)( 4)( + 2)( 2).

Hence Spec(1 K2 ) 6= Spec(2 K2 ) but (1 K2 ) = (2 K2 ) = 20.

Acknowledgement

The author thanks the University Grants Commission(India) for providing grants under minor
research project No.47-902/14 during XII plan.

References

[1] B. D. Acharya, Spectral criterion for cycle balance in networks, J. Graph Theory, 4(1980)
1 - 11.
 Spectra and Energy of Signed Graphs 21

[2] N. Alon, Eigenvalues and expanders, Combinatorica, 6(1986) 83-96.

[3] M. A. Bhat, S. Pirzada, On equienergetic signed graphs, Discrete Applied Mathematics,
189(2015), 1-7.
[4] Z. Chen, Spectra of extended double cover graphs, Czechoslovak Math.J., 54(2004) 1077-
1082.
[5] D. M. Cvetkovic, M. Doob, H. Sachs, Spectra of Graphs, Academic Press, New York, 1980.
[6] C. A. Coulson, G. R. Rushbrooke, Note on the method of molecular orbitals, Poc. Cam-
bridge Phil. Soc., 36(1940) 193-200.
[7] K. A. Germina, K. S. Hameed, On signed paths, signed cycles and their energies, Applied
Math Sci.,4(2010), 3455-3466.
[8] K. A. Germina, K. S. Hameed, T. Zaslavsky, On product and line graphs of signed graphs,
their eigenvalues and energy, Linear Algebra Appl., 435(2011), 2432-2450.
[9] M. K. Gill, B. D. Acharya, A recurrence formula for computing the characteristic polyno-
mial of a sigraph, J. Combin. Inform. Syst. Sci., 5(1) (1980), 68 - 72.
[10] R. A. Horn, C. R. Johnson, Matrix Analysis, Cambridge University Press, Cambridge,
1985.
[11] A. M. Mathai, T. Zaslavsky, On Adjacency matrices of simple signed cyclic connected
graphs, Journal of Combinatorics, Information and System Sciences, 37(2012) 369-382.
[12] N. G. Nayak, On the energy of signed graphs , Conference proceedings of International
Conference on Recent trends in Discrete Mathematics And Its Applications to Science and
Engineering, Periyar Maniammai University Tamil Nadu: Dec.2013, pp. 95-103.
[13] N. G. Nayak, Equienergetic net-regular signed graphs, International Journal of Contem-
porary Mathematical Sciences, 9(2014) 685-693. .
[14] G. Nutan, Spectra and Energy of Signed Graphs, M. Phil. Dissertation, Bharathiar Uni-
versity, Coimbatore, 2008.
[15] D. B. West, Introduction to Graph Theory, Prentice-Hall of India Pvt. Ltd., 1996.
[16] T. Zaslavsky, A mathematical bibliography of signed and gain graphs and allied areas,
(Manuscript prepared with Marge Pratt), Journal of Combinatorics, (2012), DS No. 8,
pp.1340.
Math.Combin.Book Ser. Vol.1(2017), 22-35

On Transformation and Summation Formulas for Some

Basic Hypergeometric Series

D.D.Somashekara1 , S.L.Shalini2 and K.N.Vidya1

1. Department of Studies in Mathematics, University of Mysore, Manasagangotri, Mysuru-570006, India

2. Department of Mathematics, Mysuru Royal Institute of Technology, Srirangapatna-571438, India

E-mail: [email protected], [email protected], [email protected]

Abstract: In this paper, we give an alternate and simple proofs for Sears three term
3 2 transformation formula, Jacksons 3 2 transformation formula and for a nonterminating
form of the q-Saalsch
utz sum by using q-exponential operator techniques. We also give
an alternate proof for a nonterminating form of the q-Vandermonde sum. We also obtain
some interesting special cases of all the three identities, some of which are analogous to the
identities stated by Ramanujan in his lost notebook.
Key Words: Transformation formula, q-series, operator identity.
AMS(2010): 33D15.

1. Introduction

In 1951 Sears [15] has established the following useful three term transformation formula for
3 2 series.

Theorem 1.1
 n
X (a, b, c)n ef (b, e/a, f /a, ef /bc) X (a, e/b, f /b)n n
= q
n=0
(q, e, f )n abc (e, f, b/a, ef /abc) n=0 (q, aq/b, ef /bc)n

(a, e/b, f /b, ef /ac) X (b, e/a, f /a)n n
+ q , (1.1)
(e, f, a/b, ef /abc) n=0 (q, bq/a, ef /ac)n
 
 ef 
where |q| < 1, 
  < 1 and as usual
abc 

Y
(a) := (a; q) := (1 aq n ),
n=0
(a)
(a)n := (a; q)n := , n is an integer,
(aq n )
(a1 , a2 , a3 , , am )n = (a1 )n (a2 )n (a3 )n (am )n , n is an integer or .

1 Received February 2, 2016, Accepted February 5, 2017.

 On Transformation and Summation Formulas for Some Basic Hypergeometric Series 23

Recently, Liu [9] has established (1.1) by parameter augmentation method. This formula
was used by Agarwal [1] to deduce an identity of Andrews [2, Thoerem 1] which was instrumental
in deriving sixteen partial theta function identities of Ramanujan found in his lost notebook
[4], [11].

The main objective of this paper is to give an alternate proof for (1.1) and to give proofs
for Jacksons 3 2 transformation formula and for a nonterminating form of the q-Saalsch utz
sum found in [5] by using q-exponential operator techniques. And also we give a simple proof
for a nonterminating form of the q-Vandermonde sum. Also we obtain a number of interesting
applications of these formulas.

We first list some definitions and identities that we use in the remainder of this paper. For
any function f , the q-difference operator Dq,a is defined by

f (a) f (aq)
Dq,a {f (a)} = .
a
The q-shift operator a is defined by

a {f (a)} = f (aq)

and the operator a is given by

a = 1 Dq,a .

The operator identity T (bDq,a ) [9] is defined by

X (bDq,a )n
T (bDq,a ) = (1.2)
n=0
(q; q)n

and the basic identity for T (bDq,a ) operator is

 
1 (abst; q)
T (bDq,a ) = . (1.3)
(as, at; q) (as, at, bs, bt; q)

The Cauchy operator T (a, b; Dq,c) [6] is defined by

X (a; q)n
T (a, b; Dq,c) := (bDq,c )n . (1.4)
n=0
(q; q)n

The two basic identities for the Cauchy operator (1.4) are
 
1 (abt; q)
T (a, b; Dq,c) = , |bt| < 1, (1.5)
(ct; q) (bt, ct; q)
 
(cv; q) (abs, cv; q) X (a, cs, v/t)n
T (a, b; Dq,c ) = (bt)n . (1.6)
(cs, ct; q) (bs, cs, ct; q) n=0 (q, cv, abs)n
24 D.D.Somashekara, S.L.Shalini and K.N.Vidya

The q-exponential operator R(bDq,a ) [7] is defined by

n
(1)n q (2 ) bn n
X
R(bDq,a ) = Dq,a . (1.7)
n=0
(q; q)n

The two basic identities for R(bDq,a ) are

 
1 (bt; q)
R(bDq,a ) = (1.8)
(at; q) (at; q)

and  
(av; q) (bs; q) X (v/t, b/a)n
R(bDq,a ) = (at)n . (1.9)
(at, as; q) (as; q) n=0 (q, bs)n

The q-binomial theorem [5, equation(II.3), p.354] is given by

X (a)n n (az)
z = . (1.10)
n=0
(q)n (z)

Heines transformations for 2 1 -series [5, equation(III.1), (III.2), p.359] is given by

X (, )n n (, z) X (/, z)n n
z = . (1.11)
n=0
(q, )n (, z) n=0 (q, z)n

The Rogers-Fine identity [12, equation(12), p.576] is given by

2
X ()n n X (, zq/)n n z n q n n (1 zq 2n )
z = . (1.12)
n=0
()n n=0
()n (z)n+1

The Sears transformation for 3 2 -series [5, equation (III.9), p.359] is given by
 n
X (, , )n (/, /) X (, /, /)n  n
= . (1.13)
n=0
(q, , )n (, /) n=0 (q, , /)n

The three-term 2 1 transformation formula [5, equation (III.31), p.363] is given by

 n
X (, )n n (z/, q/) X (/, q/z)n q
z =
n=0
(q, )n (z/, q/) n=0 (q, q/z)n

(, q/, /, z/q, q 2/z) X (q/, q/)n n
z . (1.14)
(/q, q/, q/, z/, q/z) n=0 (q, q 2 /)n

The Jacksons transformation [3, p. 526] is given by

X (, )n n (z) X (, /)n (z)n n(n1)/2
z = q . (1.15)
n=0
(q, )n (z) n=0 (, z, q)n
 On Transformation and Summation Formulas for Some Basic Hypergeometric Series 25

The Ramanujans [10, Ch. 16] definition of the theta function is

X
f (a, b) = an(n+1)/2 bn(n1)/2 , |ab| < 1. (1.16)
n=

The Jacobis triple product identity [8] is given by

X 2
q n z n = (qz, q/z, q 2; q 2 ) , z 6= 0. (1.17)
n=

If we set qz = a, q/z = b in (1.17), we obtain

f (a, b) = (a; ab) (b; ab) (ab; ab) , (1.18)

which is the Jacobis triple product identity in Ramanujans notation [10, Ch.16, entry 19]. It
follows from (1.16) and (1.18) that [10, Ch. 16, entry 22]

X 2 (q; q 2 ) (q 2 ; q 2 )
(q) := f (q, q) = 1 + 2 qn = , (1.19)
n=1
(q; q 2 ) (q 2 ; q 2 )

X (q 2 ; q 2 )
(q) := f (q, q 3 ) = q n(n+1)/2 = , (1.20)
n=0
(q; q 2 )

X
f (q) := f (q, q 2 ) = (1)n q n(3n1)/2 = (q; q) (1.21)
n=

and
(q) := (q; q 2 ) . (1.22)

The Ramanujans functions are given by [4], [11]

X 2
G6 (q) := (q 3 ; q 6 )2 (q 6 ; q 6 ) = (1)n q 3n = (q 3 ), (1.23)
n=

X 2
H6 (q) := (q; q 6 ) (q 5 ; q 6 ) (q 6 ; q 6 ) = (1)n q 3n 2n
= f (q, q 5 ) (1.24)
n=

and

X
J6 (q) := (q; q 3 ) (q 2 ; q 3 ) (q 3 ; q 3 ) = q n(3n+1)/2 = f (q, q 2 ). (1.25)
n=0

2. Main Theorems

In this section, we prove the main results.

Proof of Theorem 1.1. Setting = b, = a/c, = qb/c and z = q in (1.14), we obtain

26 D.D.Somashekara, S.L.Shalini and K.N.Vidya

X (b, a/c)n n (a, c/b) X (q/a)n  a n
q =
n=0
(q, qb/c)n (c, q/b) n=0 (q)n b

(a/c, c/b, b) X (c, a/b)n n
q . (2.1)
(b/c, a/b, c) n=0 (q, qc/b)n

On using q-binomial theorem for the first series on the right side of (2.1), we obtain

X (b, a/c)n n (a/c, c/b, b) X (c, a/b)n n (a, c/b)
q + q = . (2.2)
n=0
(q, qb/c)n (b/c, a/b, c) n=0
(q, qc/b)n (c, a/b)

Divide the identity (2.2) throughout by (a/c, c/b, b) to obtain

(a) 1 X (b)n q n
=
(b, c, a/b, a/c) (b, c/b) n=0 (q, qb/c)n (aq n /c)

1 X (c)n q n
+ . (2.3)
(c, b/c) n=0 (q, qc/b)n (aq n /b)

Applying T (d, e; Dq,a ) to both the sides of the identity (2.3) and using (1.5) and (1.6), we
obtain

(a, de/b) X (d, a/b, c)n  e n 1 X (b)n (deq n /c) q n
=
(b, c, a/b, a/c, e/b) n=0 (q, de/b, a)n c (b, c/b) n=0 (q, qb/c)n (aq n /c, eq n /c)

1 X (c)n (deq n /b) q n
+ . (2.4)
(c, b/c) n=0 (q, qc/b)n (aq n /b, eq n /b)

Multiply the identity (2.4) throughout by (b, c, a/b, a/c, e/b)/(a, de/b) to obtain

X (d, a/b, c)n  e n (c, a/b, e/b, de/c) X (b, a/c, e/c)n n
= q
n=0
(q, de/b, a)n c (a, c/b, e/c, de/b) n=0 (q, qb/c, de/c)n

(b, a/c) X (c, a/b, e/b)n n
+ q . (2.5)
(a, b/c) n=0 (q, qc/b, de/b)n

Change a to A, b to C, c to B, d to A/D and e to E in (2.5) to obtain

 n
X (B, A/D, A/C)n E
n=0
(q, A, AE/CD)n B

(B, A/C, E/C, AE/BD) X (C, A/B, E/B)n
= qn
(A, B/C, E/B, AE/CD) n=0 (q, Cq/B, AE/BD)n

(C, A/B) X (B, A/C, E/C)n
+ qn . (2.6)
(A, C/B) n=0 (q, Bq/C, AE/CD)n
 On Transformation and Summation Formulas for Some Basic Hypergeometric Series 27

Setting = B, = A/D, = A/C, = A and = AE/CD in (1.13), using the resulting

identity on the left side of (2.6) and then multiplying the resulting identity throughout by
(E/B, AE/CD) /(E, AE/BCD) ; change A to e, B to b, C to a, D to c and E to f in the
resulting identity, we obtain (1.1). 2
Remark 1. The identity (2.3) can be used to prove Lemma 2.1 of Somashekara, Narasimha
Murthy and Shalini [13], which played a key role in giving a unified approach to the proofs of
the reciprocity theorem of Ramanujan and its generalizations.

Remark 2. The identity (2.3) can also be used to prove Theorem 2.2 of Somashekara, Kim,
Kwon and Shalini [14], which played a key role in giving proofs for ten identities of Ramanujan
found in his lost notebook [4].

Theorem 2.1([5, equation III.5, p. 359]) We have

X (a, b)n n (abz/c) X (a, c/b, 0)n n
z = q
n=0
(q, c)n (bz/c) n=0 (q, c, cq/bz)n

(a, bz, c/b) X (z, abz/c, 0)n n
+ q . (2.7)
(c, z, c/bz) n=0 (q, bz, bzq/c)n

Proof Applying R(dDq,a ) to both the sides of the identity (2.3) and using (1.8), (1.9), we
obtain

(d/c) X (b, d/a)n  a n 1 X (b)n (dq n /c)
= qn
(b, c, a/c) n=0 (q, d/c)n b (b, c/b) n=0 (q, bq/c)n (aq n /c)

1 X (c)n (dq n /b)
+ qn . (2.8)
(c, b/c) n=0 (q, cq/b)n (aq n /b)

Multiply the identity (2.8) throughout by (b, c, a/c) /(d/c) to obtain

X (b, d/a)n  a n (c) X (b, a/c, 0)n n
= q
n=0
(q, d/c)n b (c/b) n=0 (q, bq/c, d/c)n

(b, a/c, d/b) X (c, a/b, 0)n n
+ q . (2.9)
(a/b, b/c, d/c) n=0 (q, cq/b, d/b)n

Change a to az, b to a, c to abz/c and d to abz in (2.9) to obtain (2.7). 2

Theorem 2.2([5, equation II.23, p. 356]) We have

X (a, b)n n (q/c, a, b) X (aq/c, bq/c)n n (q/c, abq/c)
q + 2
q = . (2.9)
n=0
(q, c)n (c/q, aq/c, bq/c) n=0 (q, q /c)n (aq/c, bq/c)
28 D.D.Somashekara, S.L.Shalini and K.N.Vidya

Proof Change lower case letters to upper case letters in (2.2) and then change B to a,
A/C to b and Bq/C to c to obtain (2.9). 2

Theorem 2.3([5, equation II.24, p. 356]) We have

X (a, b, c)n n (q/e, a, b, c, qf /e) X (aq/e, bq/e, cq/e)n n
q + 2
q
n=0
(q, e, f )n (e/q, aq/e, bq/e, cq/e, f ) n=0 (q, q /e, qf /e)n

(q/e, f /a, f /b, f /c)

= , (2.10)
(aq/e, bq/e, cq/e, f )

where ef = abcq.

Proof Divide (2.3) throughout by (a) to obtain

1 1 X (b)n q n
=
(b, c, a/b, a/c) (b, c/b) n=0 (q, qb/c)n (aq n /c, a)

1 X (c)n q n
+ . (2.11)
(c, b/c) n=0 (q, qc/b)n (aq n /b, a)

Applying T (dDq,a ) to both the sides of the identity (2.11) and using (1.3), we obtain

(ad/bc) 1 X (b)n (adq n /c)
= qn
(b, c, a/b, a/c, d/b, d/c) (b, c/b) n=0 (q, bq/c)n (aq n /c, a, dq n /c, d)

1 X (c)n (adq n /b)
+ qn . (2.12)
(c, b/c) n=0 (q, qc/b)n (aq n /b, a, dq n /b, d)

Multiply the identity (2.12) throughout by (a, b, d, a/c, c/b, d/c)/(ad/c) to obtain

X (b, a/c, d/c)n q n (c/b, b, a/c, d/c, ad/b) X (c, a/b, d/b)n q n
+
n=0
(q, bq/c, ad/c)n (b/c, c, a/b, d/b, ad/c) n=0 (q, qc/b, ad/b)n
(c/b, ad/bc, d, a)
= . (2.13)
(c, a/b, d/b, ad/c)

Change lower case letters to upper case letters in (2.13) and then change B to a, A/C to
b, D/C to c, Bq/C to e and AD/C to f to obtain (2.10). 2

3. Some Applications of Main Results

In this section, we derive some interesting special cases of the main identities. These special
cases are found to be analogues to some identities of Ramanujan found in his lost notebook [4],
[11].
 On Transformation and Summation Formulas for Some Basic Hypergeometric Series 29

Setting a = C, b = B/A, c = D and z = A in (2.7), we obtain

X (C, B/A)n n (BC/D) X (C, AD/B)n n
A = q
n=0
(q, D)n (B/D) n=0 (q, D, qD/B)n

(B, C, AD/B) X (A, BC/D)n n
+ q . (3.1)
(A, D, D/B) n=0 (q, B, qB/D)n

Change B to , C to , D to q and then let A 0 in (3.1) to obtain

X (1)n n q n(n1)/2 (/q) X qn
=
n=0
(q; q)n (1 q n ) (/ q) n=0 (q; q)n ( q 2 /)n (1 q n )

(1 /q)() X qn
+ . (3.2)
( q/) n=0
(q; q)n (/ )n (1 q n1 )

Change q to q 2 and set = 1 and = q 3 in (3.2) to obtain

2
X q n +2n (q; q 2 ) X q 2n
2 2 2n
=
n=0
(q ; q )n (1 + q ) (q; q ) n=0 (q; q)2n1 (1 q 4n )
2

(q; q 2 ) X q 2n+1
. (3.3)
(q; q ) n=0 (q; q)2n (1 q 4n+2 )
2

Use (1.22) to obtain

2
X q n +2n (q) X q 2n
2 2 4n
=
n=0
(q ; q )n1 (1 q ) (q) n=0 (q; q)2n1 (1 q 4n )

(q) X q 2n+1
.
(q) n=0 (q; q)2n (1 q 4n+2 )

Setting = B/A, = C, = D and z = A in (1.11), we obtain

X (B/A, C)n n (B, C) X (A, D/C)n n
A = C . (3.4)
n=0
(q, D)n (A, D) n=0 (q, B)n

Using (3.4) in (3.1) and then multiplying the resulting identity throughout by (A, D) /(B, C) ,
we obtain

X (D/C, A)n n (A, D, BC/D) X (C, AD/B)n n
C = q
n=0
(q, B)n (B, C, B/D) n=0 (q, D, Dq/B)n

(AD/B) X (A, BC/D)n n
+ q . (3.5)
(B/D) n=0 (q, B, qB/D)n

Change q to q 2 and set A = t, B = aq 3 , C = a and D = aq 2 in (3.5) and then let

30 D.D.Somashekara, S.L.Shalini and K.N.Vidya

t 0; divide the resulting identity throughout by (1 + aq), we obtain

X (1)n an 1 X q 2n
2
=
n=0
(aq; q )n+1 (q; q ) n=0 (q; q)2n (1 + aq 2n )
2

1 X q 2n+1
. (3.6)
(q; q 2 ) n=0 (q; q)2n+1 (1 + aq 2n+1 )

In Rogers-Fine identity, change q to q 2 , set = 0, = aq 3 and z = a; multiply the

resulting identity throughout by 1/(1 + aq) to obtain

2 2
X (1)n an X a2n q 2n +n X a2n q 2n +n
= =
n=0
(aq; q 2 )n+1 n=0
(a; q 2 )n+1 (aq; q 2 )n+1 n=0
(a; q)2n+2
2
X a2n q 2n +n ((1 + aq 2n+1 ) aq 2n+1 )
=
n=0
(a; q)2n+2
2 2
X a2n q 2n +n X a2n+1 q 2n +3n+1 X (1)n an q n(n+1)/2
= = . (3.7)
n=0
(a; q)2n+1 n=0 (a; q)2n+2 n=0
(a; q)n+1

Use (3.7) in (3.6) and also use (1.21) to obtain

X (1)n an q n(n+1)/2 f (q 2 ) X q 2n
=
n=0
(a; q)n+1 f (q) n=0 (q; q)2n (1 + aq 2n )

f (q 2 ) X q 2n+1
. (3.8)
f (q) n=0 (q; q)2n+1 (1 + aq 2n+1 )

Change q to q 2 , set A = t, B = aq 3 , C = aq and D = aq 3 in (3.5) and let t 0 in the

resulting identity; multiply the resulting identity throughout by 1/(1 aq) and also use (1.21)
to obtain on some simplifications

X (1)n an q n f (q 2 ) X q 2n
= . (3.9)
n=0
(aq; q 2 )n+1 f (q 4 ) n=0 (q 4 ; q 4 )n (1 a2 q 4n+2 )

In Rogers-Fine identity, replace q by q 2 , set = 0, = aq 3 and z = aq and then multiply

the resulting identity throughout by 1/(1 aq) to obtain
2
X (1)n an q n X (1)n a2n q 2n +2n
= . (3.10)
n=0
(aq; q 2 )n+1 n=0
(a2 q 2 ; q 4 )n+1

Use (3.10) in (3.9) to obtain

2
X (1)n a2n q 2n +2n f (q 2 ) X q 2n
= . (3.11)
n=0
(a2 q 2 ; q 4 )n+1 f (q 4 ) n=0 (q 4 ; q 4 )n (1 a2 q 4n+2 )

Change q to q 2 , set A = t, B = aq 3 , C = aq and D = aq 3 in (3.5) and multiply the

 On Transformation and Summation Formulas for Some Basic Hypergeometric Series 31

resulting identity throughout by 1/(1 aq) to obtain

X (t; q 2 )n (aq)n (t; q 2 ) X (t; q 2 )n q 2n
2
=
n=0
(aq; q )n+1 2(q ; q ) n=0 (q ; q 4 )n (1 + aq 2n+1 )
2 2 4

(t; q 2 ) X (t; q 2 )n q 2n
+ . (3.12)
2(q 2 ; q 2 ) n=0 (q 4 ; q 4 )n (1 aq 2n+1 )

Set a = 1 and t = q in (3.12) to obtain

X (q; q 2 )n q n (q; q 2 ) X (q; q 2 )n q 2n
=
n=0
(q; q 2 )n+1 2(q 2 ; q 2 ) n=0 (q 4 ; q 4 )n (1 q 2n+1 )

(q; q 2 ) X (q; q 2 )n q 2n
+ . (3.13)
2(q 2 ; q 2 ) n=0 (q 4 ; q 4 )n (1 + q 2n+1 )

In Rogers-Fine identity, replace q by q 2 , set = z = q and = q 3 ; multiply the resulting

identity throughout by 1/(1 + q) to obtain

X (q; q 2 )n q n X
2)
= (1)n q 2n(n+1) . (3.14)
n=0
(q; q n+1 n=0

Use (3.14) in (3.13) and also use (1.19), (1.20) and (1.21) to obtain

X f (q) X (q; q 2 )n q 2n
2 (1)n q 2n(n+1) =
n=0
f (q 4 ) n=0 (q 4 ; q 4 )n (1 q 2n+1 )

(q) X (q; q 2 )n q 2n
+ . (3.15)
(q) n=0 (q ; q 4 )n (1 + q 2n+1 )
4

In (3.5), set A = q, B = aq, C = and D = a2 q to obtain

X (a2 q/ )n n ( /a, q, a2 q) X ( )n
= 2 q)
qn
n=0
(aq) n (1/a, aq, ) n=0
(q, a n

(aq) X ( /a)n
+ qn . (3.16)
(1/a) n=0 (aq, q/a)n

In Rogers-Fine identity, set = a2 q/, = aq and z = to obtain

2
X (a2 q/ )n n X (1)n (a2 q/ )n an q n (1 a2 q 2n+1 )
= . (3.17)
n=0
(aq)n n=0
( )n+1
32 D.D.Somashekara, S.L.Shalini and K.N.Vidya

Use (3.17) in (3.16) and then let 0 in the resulting identity to obtain

X
3n n(3n+1)/2 2 2n+1 (a2 q) f 2 (q) X qn
a q (1 a q )=
n=0
f (aq, 1/a) n=0 (q, a2 q)n

(aq) X qn
+ . (3.18)
(1/a) n=0 (aq, q/a)n

Set a = 1 in (3.18) to obtain

X f 3 (q) X q n 1 X qn
q n(3n+1)/2 (1 q 2n+1 ) = + . (3.19)
n=0
f (q, 1) n=0 (q)2n 2 n=0 (q)2n

In (1.11), set = z = q and then = 0, = 0 to obtain

X qn 1 X
2
= 2
(1)n q n(n+1)/2 . (3.20)
n=0
(q)n (q)n n=0

Use (3.20) in (3.19) and also use (1.20) to obtain

X f (q)(q) 1 X q n
q n(3n+1)/2 (1 q 2n+1 ) = + .
n=0
f (q, 1) 2 n=0 (q)2n

In (3.16), let 0 to obtain

X (1)n a2n q n(n+1)/2 (q, a2 q) X qn
=
n=0
(aq)n (1/a, aq) n=0 (q, a2 q)n

(aq) X qn
+ . (3.21)
(1/a) n=0 (aq, q/a)n

Set a = 1 in (3.21) to obtain

X (1)n q n(n+1)/2 f 3 (q) X q n 1 X qn
= + . (3.22)
n=0
(q)n f (1, q) n=0 (q)2n 2 n=0 (q)2n

The left side of (3.22) yields

X (1)n q n(n+1)/2 X q n(2n+1) X q (n+1)(2n+1)
=
n=0
(q)n n=0
(q; q)2n n=0 (q; q)2n+1

X q n(2n+1) ((1 + q 2n+1 ) q 2n+1 ) X q n(2n+1)
= = . (3.23)
n=0
(q; q)2n+1 n=0
(q; q)2n+1
 On Transformation and Summation Formulas for Some Basic Hypergeometric Series 33

Use (3.23) in (3.22) to obtain

X q n(2n+1) f 3 (q) X q n 1 X qn
= + . (3.24)
n=0
(q; q)2n+1 f (1, q) n=0 (q)2n 2 n=0 (q)2n

Use (3.20) in (3.24) to obtain

X q n(2n+1) f 3 (q) X 1 X qn
= (1)n q n(n+1)/2 + . (3.25)
n=0
(q; q)2n+1 f (1, q) n=0 2 n=0 (q)2n

Use the definition of to obtain

X q n(2n+1) f (q)(q) 1 X q n
= + .
n=0
(q; q)2n+1 f (1, q) 2 n=0 (q)2n

In (3.5), replace q by q 2 , set A = q 2 , B = aq 3 , C = and D = a2 q 2 ; multiply the resulting

identity throughout by 1/(1 + aq) to obtain

X (a2 q 2 / ; q 2 )n n (q 2 , a2 q 2 , q /a; q 2 ) X ( ; q 2 )n
2)
= 2) 2 , a2 q 2 ; q 2 )
q 2n
n=0
(aq; q n (aq, , q/a; q n=0
(q n

(aq; q 2 ) X (q /a; q 2 )n
+ q 2n . (3.26)
(q/a; q 2 ) n=0 (q 3 /a; q 2 )n (aq; q 2 )n+1

In Rogers-Fine identity, replace q by q 2 , set = a2 q 2 /, = aq 3 , z = and then multiply

the resulting identity throughout by 1/(1 + aq) to obtain

X (a2 q 2 / ; q 2 )n n X (1)n (a2 q 2 / ; q 2 )n n an q 2n(n+1) (1 a2 q 4n+2 )
= . (3.27)
n=0
(aq; q 2 )n n=0
(1 + aq 2n+1 )( ; q 2 )n+1

Use (3.27) in (3.26) and then let 0 to obtain

X 2 (q 2 , a2 q 2 ; q 2 ) X q 2n
a3n q 3n +2n
(1 aq 2n+1 ) =
n=0
(q/a, aq; q 2 ) n=0 (q 2 , a2 q 2 ; q 2 )n

(aq; q 2 ) X q 2n
+ . (3.28)
(q/a; q ) n=0 (aq; q )n+1 (q 3 /a; q 2 )n
2 2

In (3.5), replace q to q 2 , set A = q 2 , B = q 3 , D = q 2 and then let C 0; multiply the

resulting identity throughout by 1/(1 + q) to obtain
2
X (1)n q n +n (q 2 ; q 2 )2 X q 2n X q 2n
= + (1 + q) . (3.29)
n=0
2
(q; q )n+1 2 2 2 2
(q; q ) n=0 (q ; q )n2
n=0
(q; q 2 )2n+1
34 D.D.Somashekara, S.L.Shalini and K.N.Vidya

In (2.10), replace q by q 6 , set a = q, b = q 4 , c = q 2 , e = q 3 and f = q 7 to obtain

X (q 2 , q 4 ; q 6 )n q 6n (q 2 , q 4 ; q 6 ) X (q; q 6 )n+1 (q 5 ; q 6 )n q 6n+3
3 6 6 6n+1

n=0
(q , q ; q )n (1 q ) (q, q ; q ) n=0 (q ; q 6 )n+1 (q 6 ; q 6 )n (1 q 6n+4 )
5 6 3

(q 3 ; q 6 )2 (q 6 ; q 6 )
= (1 q) . (3.30)
(q; q 6 )2 (q 4 ; q 6 )

Use (1.21), (1.23) and (1.24) to obtain on some simplifications

X (q 2 , q 4 ; q 6 )n q 6n f (q 2 ) X (q; q 6 )n+1 (q 5 ; q 6 )n q 6n+3

n=0
(q 3 , q 6 ; q 6 )n (1 q 6n+1 ) H6 (q) n=0 (q 3 ; q 6 )n+1 (q 6 ; q 6 )n (1 q 6n+4 )
G26 (q)H62 (q)f (q 2 )
= (1 q) . (3.31)
(q; q 6 )2 f (q)f 2 (q 6 )

In (2.10), replace q by q 3 , set a = c = q, b = e = q 2 and f = q 3 to obtain

X (q; q 3 )2n 3n (q; q 3 )2 (q 2 ; q 6 )2 X (q 2 ; q 3 )2n q 3n+1
q +
n=0
(q 3 ; q 3 )2n (q; q)2 n=0
(q; q 3 )2n+1
(q; q 3 )2 (q 2 ; q 3 )2 (q 3 ; q 3 )2
= . (3.32)
(q 2 ; q 3 )2 (q 3 ; q 3 )4

Use (1.25) to obtain

X (q; q 3 )2n 3n (q; q 3 )2 (q 2 ; q 6 )2 X (q 2 ; q 3 )2n q 3n+1 J62 (q)(q; q 3 )2
q + = .
n=0
(q 3 ; q 3 )2n (q; q)2 n=0
(q; q 3 )2n+1 (q; q)2 (q 3 ; q 3 )4

Acknowledgement

The first author is thankful to University Grants Commission(UGC), India for the financial
support under the grant SAP-DRS-1-NO.F.510/2/DRS/2011 and the second author is thank-
ful to UGC for awarding the Rajiv Gandhi National Fellowship, No.F1-17.1/2011-12/RGNF-
SC-KAR-2983/(SA-III/Website) and the third author is thankful to UGC for awarding the Ba-
sic Science Research Fellowship, No.F.25-1/2014-15(BSR)/No.F.7-349/2012(BSR). The authors
are thankful to Prof. Z.G. Liu of East China Normal University for his valuable suggestions.

References

[1] R. P. Agarwal, On the paper, A lost notebook of Ramanujan, Adv. in Math. 53(1984),
291-300.
[2] G. E. Andrews, Ramanujans lost notebook. I. partial -functions, Adv. in Math. 41(1981),
137-172.
[3] G. E. Andrews, R. Askey and R. Roy, Special Functions, Cambridge University Press,
Cambridge, 1999.
 On Transformation and Summation Formulas for Some Basic Hypergeometric Series 35

[4] G. E. Andrews and B. C. Berndt, Ramanujans Lost Notebook, Part II, Springer, New
York, 2009.
[5] G. Gasper and M. Rahman, Basic Hypergeometric Series, Encyclopedia of Mathematics,
Cambridge University press. Cambridge, 2004.
[6] W.Y.C. Chen, and N.S.S. Gu, The Cauchy operator for basic hypergeometric series, Ad-
vances in Applied Mathematics 41 (2008), 177-196.
[7] Husam L. Saad and Abaas A. Sukhi, The q-Exponential Operator, Applied Mathematical
Sciences 7(128)(2013), 6369-6380.
[8] C. G. J. Jacobi, Fundamenta Nova Theoriae Funtionum Ellipticarum, Sumptibus Fratrum
Borntrager, Regiomonti, 1829; reprinted in Gesamelte Werke, 1, Reimer, Berlin, 1881,
49-239; reprinted by Chelsea, New York, 1969.
[9] Z.-G. Liu, Some operator identities and q-series transformation formulas, Discrete Mathe-
matics 265(2003), 119-139.
[10] S. Ramanujan, Notebooks (2 Volumes), Tata Institute of Fundamental Research, Bombay,
1957.
[11] S. Ramanujan, The Lost Notebook and Other Unpublished Papers, Narosa, New Delhi,
1988.
[12] L. J. Rogers, Second memior on the expansion of certain infinite products, Proc. London
Math. Soc. 25(1894), 318-343.
[13] D. D. Somashekara, K. Narasimha murthy and S. L. Shalini, On the reciprocity theorem
of Ramanujan and its generalizations, Proc. Jangjeon Math. Soc. 15(3)(2012), 343-353.
[14] D. D. Somashekara, T. Kim, H. I. Kwon and S. L. Shalini, On some partial theta function
identities of Ramanujan found in his lost notebook, Adv. Studies Contemp. Math., 25 (1),
2015.
[15] D.B. Sears, On the transformation theory of basic hypergeometric functions, Proc. London
Math. Soc., 53 (1951), 158-180.
Math.Combin.Book Ser. Vol.1(2017), 36-52

Some New Generalizations of the Lucas Sequence

F
ugen TORUNBALCI AYDIN
Yildiz Technical University, Faculty of Chemical and Metallurgical Engineering,

Department of Mathematical Engineering, 34220, Istanbul, Turkey

Salim YUCE
Yildiz Technical University, Faculty of Arts and Sciences

Department of Mathematics, 34220, Istanbul, Turkey

E-mail: [email protected], [email protected]

Abstract: In this paper, we investigate the generalized Lucas, the generalized complex
Lucas and the generalized dual Lucas sequence using the Lucas number. Also, we investigate
special cases of these sequences. Furthermore, we give recurrence relations, vectors, the
golden ratio and Binets formula for the generalized Lucas and the generalized dual Lucas
sequence.

Key Words: Smarandache-Fibonacci triple, Fibonacci number, Lucas number, Lucas

sequence, generalized Fibonacci sequence, generalized complex Lucas sequence, generalized
dual Lucas sequence.

AMS(2010): 11B37, 11B39.

1. Introduction

Let S(n), n 0 with S(n) = S(n 1) + S(n 2) be a Smarandache-Fibonacci triple, where

S(n) is the Smarandache function for integers n 0. Particularly, let S(n) be F (n) or L(n),
we get the Fibonacci or Lucas sequence as follows:

A Fibonacci sequence

1 , 1 , 2, 3, 5, 8, 13, 21, 34, 55, 89, 144, 233, , Fn ,

is defined by the recurrence relation

Fn = Fn1 + Fn2 , (n > 3),

with F1 = F2 = 1 , where Fn is the n-th term of the Fibonacci sequence (Fn ) (Leonardo
Fibonacci, 1202). The Fibonacci sequence is named after Italian mathematician Leonardo of
Pisa, known as Fibonacci. The name Fibonacci Sequnce was first used by the 19th- century
number theorist Edouard Lucas. Some recent generalizations for the Fibonacci sequence have
1 Received May 17, 2016, Accepted February 8, 2017.
 Some New Generalizations of the Lucas Sequence 37

produced a variety of new and extended results, [1],[5],[6],[9],[13].

A Lucas sequence

2 , 1 , 3, 4, 7, 11, 18, 29, 47, 76, 123, 199, 322, , Ln ,

is defined by the recurrence relation

Ln = Ln1 + Ln2 , (n > 3),

with L1 = 2 , L2 = 1 , where Ln is the n-th term of the Lucas sequence (Ln ) (Francois Edouard
Anatole Lucas, 1876). There are a lot of generalizations of the Lucas sequences,[15],[16],[17].

The generalized Fibonacci sequence defined by

Hn = Hn1 + Hn2 , (n > 3) (1.1)

with H1 = p , H2 = p+q where p, q are arbitrary integers [3]. That is, the generalized Fibonacci
sequence is

p , p + q , 2p + q, 3p + 2q, 5p + 3q, 8p + 5q, , (p q)Fn + qFn+1 , (1.2)

Using the equations (1.1) and (1.2) , it was obtained

Hn+1 = q Fn + p Fn+1
(1.3)
Hn+2 = p Fn + (p + q) Fn+1 .

For the generalized Fibonacci sequence, it was obtained the following properties:

2
Hn1 + Hn2 = (2p q)H2n1 e F2n1 , (1.4)

2 2
Hn+1 Hn1 = (2p q)H2n e F2n , (1.5)

Hn1 Hn+1 Hn2 = (1)n e , (1.6)

Hn+r = Hn1 Fr + Hn Fr+1 (n > 3) (1.7)

2
Hn+1r Hn+1+r Hn+1 = (1)nr e Fr2 , (1.8)
2
Hn+1 + e Fn2 = p H2n+1 , (1.9)

Hn Hn+1+r Hns Hn+r+s+1 = (1)n+s e Fs Fr+s+1 , (1.10)

[2Hn+1 Hn+2 ] 2 + [Hn Hn+3 ] 2 = 2Hn+1 Hn+2 + Hn2 2

 
(1.11)
r
Hn+r + (1) Hnr
= Fr+1 + (1)r Fr1 (1.12)
Hn
where e = p2 p q q 2 .
38 F
ugen TORUNBALCI AYDIN and Salim YUCE

Also, for p = 1, q = 0 , we get the following well-known results:

2
Fn1 + Fn2 = F2n1 , (Catalan), (1.13)

Fn1 Fn+1 Fn2 = (1)n , (Simpson or Cassini), (1.14)

2
Fn+1 + Fn2 = F2n+1 (Lucas). (1.15)

In this paper, we will define the generalized Lucas, the generalized complex Lucas and the
generalized dual Lucas sequences respectively, denoted by Gn , Cn , Dn .

2. Generalized Lucas Sequence and Lucas Vectors

In this section, we will define the generalized Lucas sequence denoted by Ln . The generalized
Lucas sequence defined by
Ln = Ln1 + Ln2 , (n > 3), (2.1)

with L1 = 2 p q , L2 = p + 2 q where p, q are arbitrary integers,[3]. That is, the generalized

Lucas sequence is

2 p q , p + 2 q , 3 p + q , 4 p + 3 q , 7 p + 4 q , 11 p + 7 q , , (p q)Ln + qLn+1 , (2.2)

Using the equations (2.1) and (2.2) , we get

Ln+1 = q Ln + p Ln+1 ,
(2.3)
Ln+2 = p Ln + (p + q) Ln+1 .

Putting n = r in (2.3) and using (2.1), we find in turn

Lr+3 = (2 p + q )Lr+1 + (p + q ) Lr = H3 Lr+1 + H2 Lr

(2.4)
Lr+4 = (3 p + 2 q) Lr+1 + (2 p + q) Lr = H4 Lr+1 + H3 Lr

So, in general, we have obtain relations between generalized Lucas sequence and generalized
Fibonacci sequence as follows:

Ln+r = Hn1 Lr + Hn Lr+1 (2.5)

Also, certain results follow almost immediately from (2.1)

Ln+2 2 Ln Ln1 = 0, (2.6)

Ln+1 2 Ln + Ln2 = 0, (2.7)

 Some New Generalizations of the Lucas Sequence 39

n1
X
L2i+1 = L2n (2 p q), (2.8)
i=0
n
X
L2i = L2n+1 (p + 2 q), (2.9)
i=1
n
X
(L2i1 L2i ) = L2n1 p + 3 q. (2.10)
i=1

For the generalized Lucas sequence, we have the following properties:

L2n1 + L2n = (2p q)(L2n2 + L2n ) eL (L2n2 + L2n ) , (2.11)

L2n+1 L2n1 = (2p q)(L2n+2 L2n2 ) eL (L2n+2 L2n2 ) , (2.12)

Ln1 Ln+1 L2n = 5 (1)n+1 eL , (2.13)

L2n+1 + eL L2n = p (L2n+2 + L2n ) , (2.14)

Ln+r + (1)r Lnr

= Lr (2.15)
Ln
where eL = p2 p q q 2 .

Theorem 2.1 If Ln is the generalized Lucas number, then

Ln+1 p + q
lim = ,
n Ln q + (p q)

where = (1 + 5) 2 = 1.618033 is the golden ratio.

Proof We have for the Lucas number Ln ,

Ln+1
lim = ,
n Ln

where
.
= (1 + 5) 2 = 1.618033

is the golden ratio [12].

Then for the generalized Lucas number Ln , we obtain

Ln+1 pLn+1 + qLn p + q

lim = lim = . (2.16)
n Ln n qLn+1 + (p q)Ln q + (p q)
2
40 F
ugen TORUNBALCI AYDIN and Salim YUCE

Theorem 2.2 The Binets formula2 for the generalized Lucas sequence is as follows;

n + n )
Ln = ( (2.17)

where = (2 p q) (p + 2 q), = (p + 2 q) (2 p q).

Proof The characteristic equation of recurrence relation Ln+2 = Ln+1 + Ln is

t2 t 1 = 0. (2.18)

The roots of this equation are

1+ 5 1 5
= and = , (2.19)
2 2

where + = 1 , = 5 , = 1.
Using recurrence relation and initial values L0 = (2 p q), L1 = (p + 2 q) the Binets
formula for Ln , we get
Ln = A n + B n = n + n ,
 
(2.20)

where
L1 L0 L0 L1
A= , B=

= (2 p q) (p + 2 q), = (p + 2 q) (2 p q).
and 2
A generalized Lucas vector is defined by


Ln = (Ln , Ln+1 , Ln+2 )

Also, from equation (2.2) it can be expressed as

Ln = (p q) L n + q L n+1 (2.21)

where L n = ( Ln , Ln+1 , Ln+2 ) and L n+1 = ( Ln+1 , Ln+2 , Ln+3 ) are the Lucas vectors.


The product of Ln and R is given by



Ln = ( Ln , Ln+1 , Ln+2 )

2 Binets formula is the explicit formula to obtain the n-th Fibonacci and Lucas numbers. It is well known

that for the Fibonacci and Lucas numbers, Binets formulas are
n n
Fn =

and
Ln = n + n
.
respectively, where + = 1 , = 5 , = 1 and = (1 + 5) 2 ,
.
= (1 5) 2 ,[7], [8].
 Some New Generalizations of the Lucas Sequence 41

and Ln and Lm are equal if and only if

Ln = Lm
Ln+1 = Lm+1
Ln+2 = Lm+2 .

Theorem 2.3 Let Ln and Lm be two generalized Lucas vectors. The dot product of Ln and Lm
is given by
D
E
Ln , Lm = p2 (5 Fn+m+3 + Ln Lm )
+p q [2 Ln+m1 + 10 Fn+m+2 ]
+q 2 (5 Fn+m+1 + Ln1 Lm1 ) . (2.22)

Proof The dot product of Ln = (Ln , Ln+1 , Ln+2 ) and

Lm = (Lm , Lm+1 , Lm+2 ) defined by

E
D
Ln , Lm = Ln Lm + Ln+1 Lm+1 + Ln+2 Lm+2 .

Also, using the equations (2.1), (2.2) and (2.3), we obtain

Ln Lm = p2 (Ln Lm ) + p q [Ln Lm1 + Ln1 Lm ] + q 2 (Ln1 Lm1 ), (2.23)

Ln+1 Lm+1 = p2 (Ln+1 Lm+1 ) + p q [Ln+1 Lm + Ln Lm+1 ] + q 2 (Ln Lm ), (2.24)

Ln+2 Lm+2 = p2 (Ln+2 Lm+2 ) + p q [Ln+2 Lm+1 + Ln+1 Lm+2 ]

(2.25)
+q 2 (Ln+1 Lm+1 ).

Then, from the equations (2.23), (2.24) and (2.25), we have

D
E
Ln , Lm = p2 (Ln Lm + Ln+1 Lm+1 + Ln+2 Lm+2 )
+(p q) [Ln Lm1 + Ln1 Lm + Ln+1 Lm + Ln Lm+1
+Ln+2 Lm+1 + Ln+1 Lm+2 ]
+q 2 (Ln1 Lm1 + Ln Lm + Ln+1 Lm+1 ) (2.26)
= p2 (5 Fn+m+3 + Ln Lm )
+(p q) [10Fn+m+2 + 2 Ln+m1 ]
+q 2 (5 Fn+m+1 + Ln1 Lm1 ) .

2
42 F
ugen TORUNBALCI AYDIN and Salim YUCE

Case 1. For the dot product of the generalized Lucas vectors Ln and L n+1 , we get
D
E
Ln , Ln+1 = Ln Ln+1 + Ln+1 Ln+2 + Ln+2 Ln+3
= p2 [ 5 F2n+4 + Ln Ln+1 ]
(2.27)
+(p q) [ 10 F2n+3 + 2 L2n ]
+q 2 [ 5 F2n+2 + Ln1 Ln ]

and D
E
Ln , Ln = (Ln )2 + (Ln+1 )2 + (Ln+2 )2
= p2 [ L2n + L2n+1 + L2n+2 ]
(2.28)
+(p q)[ 2 Ln Ln1 + 2 Ln+1 Ln + 2 Ln+2 Ln+1 ]
+q 2 [ L2n1 + L2n + L2n+1 ].

Then for the norm of the generalized Lucas vector, using identities of the Fibonacci numbers

L2n+1 + L2n = 5 F2n+1

L2n+1 L2n1 = 5 F2n
L2n+1 L2n = Ln1 Ln+2
Ln Lm + Ln+1 Lm+1 = 5 Fn+m+1

we have
2
D
E
Ln = Ln , Ln = Ln 2 + L2n+1 + L2n+2
= p2 5 F2n+3 + L2n
 

+ (p q) [2 F2n+2 + 2 Ln Ln1 ] (2.29)

+q 2 5 F2n+1 + L2n1 .
 

Case 2. For p = 1, q = 0, in the equations (2.26), (2.27) and (2.29), we have

D E
Ln , Lm = [ 5 Fn+m+3 + Ln Lm ] ,
D
E
Ln , Ln+1 = [ 5 F2n+4 + Ln Ln+1 ]

and
p

Ln = 5 F2n+3 + L2n .

Theorem 2.4 Let Ln and Lm be two generalized Lucas vectors. The cross product of Ln and

Lm is given by

Ln Lm = 5 (1)n Fmn (p2 p q q 2 ) (i + j k). (2.30)
 Some New Generalizations of the Lucas Sequence 43

Proof The cross product of Ln Lm defined by
 
 i j k 


 
Ln Lm =  Ln Ln+1 Ln+2 
 
Lm Lm+1 Lm+2 
(2.31)

= i (Ln+1 Lm+2 Ln+2 Lm+1 )

+j (Ln+2 Lm Ln Lm+2 ) + k (Ln Lm+1 Ln+1 Lm ) .

Now, we calculate the cross products. Using the property Ln Lm+1 Ln+1 Lm = 5 (1)n Fmn
we get

Ln+1 Lm+2 Ln+2 Lm+1 = 5 (1)n Fmn (p2 p q q 2 ) = 5 (1)n Fmn eL , (2.32)

Ln+2 Lm Ln Lm+2 = 5 (1)n Fmn (p2 pq q 2 ) = 5 (1)n Fmn eL , (2.33)

and

Ln Lm+1 Ln+1 Lm = 5 (1)n+1 Fmn (p2 p qq 2 ) = 5 (1)n+1 Fmn eL .

(2.34)
Then from the equations (2.32), (2.33) and (2.34), we obtain the equation (2.30).

Case 3. For p = 1, q = 0, in the equation (2.30), we have

Ln Lm = 5 (1)n Fmn (i + j k).

2



Theorem 2.5 Let Ln , Lm and Lk be the generalized Lucas vectors. The mixed product of these
vectors is D E
Ln Lm , Lk = 0. (2.35)

Proof Using Lk = (Lk , Lk+1 , Lk+2 ), we can write,
 
L Ln+1 Ln+2 
D  n
E  
Ln Lm , Lk = Lm Lm+1 Lm+2 
 
 Lk Lk+1 Lk+2 
(2.36)
= Ln (Lm+1 Lk+2 Lm+2 Lk+1 )
+Ln+1 (Lm+2 Lk Lm Lk+2 ) + Ln+2 (Lm Lk+1 Lm+1 Lk ) .
44 F
ugen TORUNBALCI AYDIN and Salim YUCE

Also, using the equations (2.32), (2.33) and (2.34), we obtain

Ln (Lm+1 Lk+2 Lm+2 Lk+1 ) + Ln+1 (Lm+2 Lk Lm Lk+2 )

+ Ln+2 (Lm Lk+1 Lk Lm+1 )
(2.37)
= 5 (1)m Fkm eL ( Ln + Ln+1 Ln+2 )
= 5 (1)m Fkm eL ( Ln+2 Ln+2 ) = 0.

Thus, we have the equation (2.35). 2

3. Generalized Complex Lucas Sequence

In this section, we will define the generalized complex Lucas sequence denoted by Cn . The
generalized complex Lucas sequence defined by

Cn = Ln + i Ln+1 , (3.1)

with C0 = (2 p q) + i (p + 2 q), C1 = (p + 2 q) + i (3 p + q), C2 = (3 p + q) + i (4 p + 3 q),

where p, q are arbitrary integers. That is, the generalized complex Lucas sequence is

(2 p q) + i (p + 2 q) , (p + 2 q) + i (3 p + q), (3 p + q) + i (4 p + 3 q) ,
(3.2)
(4 p + 3 q) + i (7 p + 4 q) , . . . , ( p q + i q )Ln + ( q + i p )Ln+1 , . . .

Case 1. From the generalized complex Lucas sequence (Cn ) for p = 1, q = 0 in the equation
(3.2), we obtain complex Lucas sequence (Cn ) as follows:

(Cn ) : 2 + i , 1 + i 3 , 3 + i 4 , 4 + i 7, , Ln + i Ln+1 , .

For the generalized complex Lucas sequence, we have the following properties:

C2n + C2n1 = [ (2p q) + i (p + 2 q) ] (C2 n2 + C2 n )

(3.3)
( 2 + i ) eL (L2 n2 + L2 n ) ,

C2n+1 + C2n1 = [ (2p q) + i (p + 2 q) ] (C2 n+2 + C2 n2 )

(3.4)
( 2 + i ) eL (L2 n+2 + L2 n2 ) ,

Cn1 Cn+1 C2n = 5 (1)n+1 ( 2 + i ) eL , (3.5)

C2n+1 + ( 2 + i ) eL L2n = [ ( 2 p + q ) + (C2 n+2 + C2 n ) , (3.6)

 Some New Generalizations of the Lucas Sequence 45

Cn+r + (1)r Cnr

= Lr . (3.7)
Cn
where eC = ( 2 + i ) eL .

4. Generalized Dual Lucas Sequence

In this section, we will define the generalized dual Lucas sequence denoted by DL
n . The gener-
alized dual Lucas sequence defined by

DL
n = Ln + Ln+1 , (4.1)

with DL L
0 = (2 p q) + ( p + 2 q) , D1 = (p + 2 q) + (3 p + q) where p, q are arbitrary integers.
That is, the generalized dual Lucas sequence is

(2 p q) + (3 p + q) , (p + 2 q) + (3 p + q), (3 p + q) + (4 p + 3 q),
( 4 p + 3 q ) + (7 p + 4 q), ( 7 p + 4 q ) + (11 p + 7 q ), (4.2)
, ( p q + q )Ln + ( q + p ) Ln+1 ,

Using the equations (4.1) and (4.2) , we get

DL
n = ( p q + q ) Ln + ( q + p ) Ln+1 ,

DL
n+1 = ( q + p ) Ln + [ p + ( p + q ) ] Ln+1 , (4.3)

DL
n+2 = [ p + ( p + q ) ] Ln + [ (p + q) + ( 2 p + q ) ]Ln+1 .

Case 1. From the generalized dual Lucas sequence (DL n ) for p = 1, q = 0 in the equation
(4.2), we obtain dual Lucas sequence (DnL ) as follows:

(DnL ) : 2 + , 1 + 3 , 3 + 4 , 4 + 7 , 7 + 11 , 11 + 18 , , Ln + Ln+1 ,

For the generalized dual Lucas sequence, we have the following properties:

(DL 2 L
n ) + (Dn1 )
2
= [ (2p q) + (p + 2 q) ] DL L
2n2 + D2n
(4.4)
eD (L2n2 + L2n ) ,

(DL 2 L 2
n+1 ) (Dn ) = [ (2p q) + (p + 2 q) ] DL L
2n+2 + D2n2
(4.5)
eD (L2n+2 L2n2 ) ,

2
(DL 2 L L
n+1 ) + eD Ln = [ p + (p + q) ] D 2n+2 + D2n , (4.6)
46 F
ugen TORUNBALCI AYDIN and Salim YUCE

DL L L 2
n1 Dn+1 (Dn ) = 5 (1)
n+1
eD , (4.7)

r L
DL
n+r + (1) Dnr
= Lr , (4.8)
DL
n

where eD = ( 1 + ) eL .

Case 2. From properties of the generalized dual Lucas sequence (DL n ) for p = 1, q = 0 in the
equations (4.4) - (4.8), we obtain dual Lucas sequence (DnL ) as follows:

(DnL )2 + (Dn1
L
)2 = (2 + ) D2n2
L L
+ D2n (1 + ) (L2n2 + L2n ) , (4.9)

L
(Dn+1 )2 (DnL )2 = (2 + ) D2n+2
L L
+ D2n2 (1 + ) (L2n+2 L2n2 ) , (4.10)

L
(Dn+1 )2 + (1 + ) Ln 2 = (1 + ) (D2n+2
L L
+ D2n ), (4.11)

L L
Dn1 Dn+1 (DnL )2 = 5 (1)n+1 (1 + ) , (4.12)

L
Dn+r + (1)r Dnr
L

L
= Lr , (4.13)
Dn

Theorem 4.1 If DL
n is the generalized dual Lucas number, then

DL
n+1 ( p q) 2 + ( p2 p q + q 2 ) + ( p q q 2 )
lim L
= ,
n Dn q 2 2 + 2 q (p q ) + (p q)2

where = 1.618033

Proof For the generalized dual Lucas number DL

n , we obtain

DL
n+1 (p q + q)Ln+1 + (q + p)Ln+2
lim = lim
n DL
n (p q + q)Ln + (q + p)Ln+1
n

(p2 pq + q 2 )Ln Ln+1 + (pq q 2 )L2n + pqL2n+1

= lim
n q 2 L2n+1 + 2q(p q)Ln Ln+1 + (p q)2 L2n
5(1)n (p2 pq q 2 )
+ lim
n q 2 L2n+1 + 2q(p q)Ln Ln+1 + (p q)2 L2n
(pq)2 + (p2 pq + q 2 ) + (pq q 2 )
= , (4.14)
q 2 2 + 2q(p q) + (p q)2

where Ln+2 = Ln+1 + Ln .

 Some New Generalizations of the Lucas Sequence 47

Case 3. For p = 1, q = 0 in the equation (4.14), we obtain

DL L
Dn+1
lim n+1
n DL
= lim
n Dn L
= + 0 = . 2
n

Theorem 4.2 The Binets formula for the generalized dual Lucas sequence is as follows:

DL n + n )
n = ( (4.15)

= ( p q + q ) + (q + p ) and = ( p q + q ) + (q + p ).
where

Proof If we use definition of the generalized dual Lucas sequence and substitute first
equation in footnote, then we get

DL
n = ( p q + q ) Ln + ( q + p )Ln+1
= ( p q + q ) (n + n ) + ( q + p ) (n+1 + n+1 )
(4.16)
= n ( p q + q + q + p ) + n (p q + q + q + p )
= n + n .

= ( p q + q ) + (q + p ) and = ( p q + q ) + (q + p ).
where 2

5. Generalized Dual Lucas Vectors

A generalized dual Lucas vector is defined by

DLn = (DL L L
n , Dn+1 , Dn+2 )

Also, from equations (4.1), (4.2) and (4.3) it can be expressed as

DLn = L n + Ln+1


(5.1)
= (p q + q) L n + (q + p ) L n+1

where Ln = ( Ln , Ln+1 , Ln+2 ) and L n = ( Ln , Ln+1 , Ln+2 ) are the generalized Lucas vector
and the Lucas vector, respectively.


The product of DLn and R is given by



DLn = Ln + Ln+1


and DLn and DLm are equal if and only if

Ln = Lm
Ln+1 = Lm+1
Ln+2 = Lm+2
48 F
ugen TORUNBALCI AYDIN and Salim YUCE

Some examples of the generalized dual Lucas vectors can be given easily as:


DL1 = (DL L L
1 , D2 , D3 )

= (L1 , L2 , L3 ) + (L2 , L3 , L4 )
= (p + 2q) + (3p + q), (3p + q) + (4p + 3q), (4p + 3q) + (7p + 4q))

DL2 = (L2 , L3 , L4 ) + (L3 , L4 , L5 )
= ((3p + q) + (4p + 3q), (4p + 3q) + (7p + 4q), (7p + 4q) + (11p + 18q))

Theorem 5.1 Let DLn and DLm be two generalized dual Lucas vectors. The dot product of DLn

and DLm is given by
D
E
DLn , DLm = p2 [(Ln Lm + 5 Fn+m+3 ) + (Ln Lm+1 + Ln+1 Lm + 10 Fn+m+4 ) ]
+p q [(5 Ln+m + 10 Fn+m+2 )
+(Ln1 Lm + Ln Lm1 + 10 Fn+m1 + 20 Fn+m+3 ) ] (5.2)
2
+q [(Ln1 Lm1 + 5 Fn+m+1 )
+(Ln1 Lm + Ln Lm1 + 10 Fn+m+2 ) ]

Proof The dot product of DLn = (DL L L
n , Dn+1 , Dn+2 ) and

DLm = (DL L L
m , Dm+1 , Dm+2 ) defined by

D
E
DLn , DLm = DL L L L L L
n Dm + Dn+1 Dm+1 + Dn+2 Dm+2
D
E D
E D E
= Ln , Lm + [ Ln , Lm+1 + Ln+1 , Lm ]

where Ln = ( Ln , Ln+1 , Ln+2 ) is the generalized Lucas vector. Also, the equations (2.1), (2.2)
and (2.3), we obtain
D
E
Ln , Lm = p2 (Ln Lm + 5 Fn+m+3 )
+ p q (5 Fn+m + 10 Fn+m+2 ) (5.3)
+q 2 (Ln1 Lm1 + 5 Fn+m+1 )
D
E
Ln , Lm+1 = p2 (Ln Lm+1 + 5 Fn+m+4 )
+ p q (5 Fn+m1 + 10 Fn+m+3 + Ln1 Lm ) (5.4)
2
+q (Ln1 Lm + 5 Fn+m+2 ) ,

and D E
Ln+1 , Lm = p2 (Ln+1 Lm + 5 Fn+m+4 )
+ p q (5 Fn+m1 + 10 Fn+m+3 + Ln Lm1 ) (5.5)
+q 2 (Ln Lm1 + 5 Fn+m+2 )
 Some New Generalizations of the Lucas Sequence 49

Then from equation (5.3), (5.4) and (5.5), we have the equation (5.2). 2


Case 1. For the dot product of generalized dual Lucas vectors DLn and DLn+1 , we get
D
E
DLn , DLn+1 = DL L L L L L
n Dn+1 + Dn+1 Dn+2 + Dn+2 Dn+3
D
E D E D E
= Ln , Ln+1 + { Ln , Ln+2 + Ln+1 , Ln+1 }
= p2 [(Ln Ln+1 + 5 F2n+4 )
+(Ln Ln+2 + Ln+1 Ln+1 + 10 F2n+5 ) ]
(5.6)
+p q [(5 Ln Ln + Ln1 Ln+1 + 10 F2n+3 )
+(Ln+1 Ln+2 + 5 F2n + 10 F2n+4 ) ]
+q 2 [(Ln1 Ln + 5 F2n+2 )
+(Ln1 Ln+1 + Ln Ln + 10 F2n+3 ) ]

and D
E
DLn , DLn = (DL 2 L 2 L
n ) + (Dn+1 ) + (Dn+2 )
2
D E D E
= Ln , Ln + 2 Ln , Ln+1
= p2 [(Ln Ln + 5 F2n+3 )
+2 (Ln Ln+1 + 5 F2n+4 ) ]
(5.7)
+p q [(5 F2n + 10 F2n+2 )
+2 (Ln Ln + Ln1 Ln+1 + 10 F2n+3 ) ]
+q 2 [(Ln1 Ln1 + 5 F2n+1 )
+2 (Ln1 Ln + 5 F2n+2 ) ] .

Then for the norm of the generalized dual Lucas vector 3 , we have


Ei q
rhD
L
DLn , DLn = (DL L L

Dn = 2 2 2
n ) + (Dn+1 ) + (Dn+2 )
p
= p2 (Ln Ln + 5 F2n+3 ) + p q(5 F2n + 10 F2n+2 )
p
+ q 2 (Ln1 Ln1 + 5 F2n+1 )] (5.8)
p
+ 2{p2(Ln Ln+1 + 5 F2n+4 ) + pq(Ln Ln + Ln1 Ln+1 + 10 F2n+3 )
p
+ q 2 (Ln1 Ln + 5 F2n+2 )}.

Case 2. For p = 1, q = 0, in the equations (5.2), (5.6) and (5.8), we have

 
L L
Dn , Dm = [(Ln Lm + 5 Fn+m+3 ) + (Ln Lm+1 + Ln+1 Lm + 10 Fn+m+4 ) ],

3 Norm of dual number as follows ([2], [14]):

1
A = a + a = a + a , A = a + a
2 a
50 F
ugen TORUNBALCI AYDIN and Salim YUCE

L
 
L
Dn , Dn+1 = [(Ln Ln+1 + 5 F2n+4 ) + (Ln Ln+2 + Ln+1 Ln+1 + 10 F2n+5 ) ]

and


p
DnL =

(Ln Ln + 5 F2n+3 ) + 2 (Ln Ln+1 + 5 F2n+4 )

(Ln Ln+1 + 5 F2n+4 )

= (Ln Ln + 5 F2n+3 ) + p .
(Ln Ln + 5 F2n+3 )

Theorem 5.2 Let DLn and DLm be two generalized dual Lucas vectors. The cross product of DLn

and DLm is given by


DLn DLm = 5 (1)n+1 Fmn (1 + ) eL (i + j k). (5.9)

Proof The cross product of DLn = Ln + Ln+1 and DLm = Lm + Lm+1 defined by




DLn DLm = (Ln Lm ) + (Ln Lm+1 + Ln+1 Lm )


where Ln is the generalized Lucas vector and Ln Lm is the cross product for the generalized


Lucas vectors Ln and Lm .


Now, we calculate the cross products Ln Lm , Ln Lm+1 and

Ln+1 Lm :
Using the property Ln Lm+1 Ln+1 Lm = 5 (1)n Fmn , we get


Ln Lm = 5 (1)n+1 Fmn (i + j k) eL , (5.10)


Ln Lm+1 = 5 (1)n+1 Fmn+1 (i + j k) eL , (5.11)

and

Ln+1 Lm = 5 (1)n+2 Fmn1 (i + j k) eL . (5.12)

Then from the equations (5.10), (5.11) and (5.12), we obtain the equation (5.9). 2
Case 3. For p = 1, q = 0 in the equations (5.9), we have

DnL Dm
L
= 5 (1)n+1 Fmn (1 + ) (i + j k).



Theorem 5.3 Let DLn , DLm and DLk be the generalized dual Lucas vectors. The mixed product
of these vectors is


 
L L L
Dn Dm , Dk = 0. (5.13)

Proof Using the properties

DLn DLm = ( Ln Lm ) + ( Ln Lm+1 + Ln+1 Lm )
 Some New Generalizations of the Lucas Sequence 51

and


DLk = Lk + Lk+1 ,

we can write,
D
E D
E D
E
DLn DLm , DLk = Ln Lm , Lk + [ Ln Lm , Lk+1
D E D E
+ Ln Lm+1 , Lk + Ln+1 Lm , Lk+1 ] .

Then using equations (5.10), (5.11) and (5.12), we obtain

D
E
(i + j k), Lk = Lk + Lk+1 Lk+2 = 0,
D E
(i + j k), Lk+1 = Lk+1 + Lk+2 Lk+3 = 0 .

Thus, we have the equation (5.13). 2

References

[1] Berzsenyi G., Sums of product of generalized Fibonacci numbers, The Fibonacci Quarterly,
13 (4), (1975), 343-344.
[2] Ercan Z. and Y uce S., On properties of the Dual quaternions, Europen Journal of Pure
and Applied Mathematics, Vol. 4, No.2 (2011) 142-146.
[3] Horadam A. F., A generalized Fibonacci sequence, American Math. Monthly, 68, (1961),
455-459.
[4] Horadam A. F., Complex Fibonacci numbers and Fibonacci quaternions, American Math.
Monthly, 70, (1963), 289-291.
[5] Horadam A. F., Basic properties of a certain generalized sequence of numbers, The Fi-
bonacci Quarterly, 3(3), (1965), 161-176.
[6] Iyer M.R., Identities involving generalized Fibonacci numbers, The Fibonacci Quarterly,
7(1), (1969), 66-73.
[7] Kalman D. And Mena R., The Fibonacci Numbers-Exposed, 76(3), (2003), 167-181.
[8] Rosen K. H., Discrete Mathematics and its Applications, McGraw-Hill 1999.
[9] Walton J. E. and Horadam A. F., Some Aspects of Generalized Fibonacci Numbers, 12(3),
(1974), 241-250.
[10] Guven I. A. and Nurkan S. K., A new approach to Fibonacci, Lucas numbers and dual
vectors, Adv. Appl. Clifford Algebras, 12(3), (2014), 241-250 .
[11] Vajda S., Fibonacci and Lucas Numbers the Golden Section, Ellis Horrowood Limited Publ.,
England, 1989.
[12] Koshy T., Fibonacci and Lucas Numbers with Applications, John Wiley and Sons, Proc.,
New York-Toronto, 2001.
[13] Y
uce S. and Torunbalc Aydn F., Generalized dual Fibonacci sequence, The International
Journal of Science & Technoledge, 4(9), (Sep 2016), 193-200.
[14] Agrawal O. P., Hamilton operators and dual number quaternions in spectral kinematik,
52 F
ugen TORUNBALCI AYDIN and Salim YUCE

Mech. Mach. Theory, Vol.22, no.6, (1987), 569-575.

[15] Siar Z. and Keskin R., The square terms in generalized Lucas sequences, Mathematika,
60(1), (Jan 2014), 85-100.
[16] Siar Z. and Keskin R., The square terms in generalized Lucas Sequence with parameters
p and q, Mathematica Scandinavica, 118(1), (2016), 13-26.
[17] Bilgici G., Two generalizations of Lucas sequence, Applied Mathematics and Computation,
245, (Oct 2014), 526-538.
Math.Combin.Book Ser. Vol.1(2017), 53-62

Fixed Point Results Under Generalized Contraction

Involving Rational Expression in Complex Valued Metric Spaces

G. S. Saluja
(Department of Mathematics, Govt. Nagarjuna P.G. College of Science, Raipur - 492010 (C.G.), India)

E-mail: [email protected]

Abstract: The aim of this paper is to study common fixed point under generalized contrac-
tion involving rational expression in the setting of complex valued metric spaces. The results
presented in this paper extend and generalize several results from the existing literature.

Key Words: Common fixed point, generalized contraction involving rational expression,
complex valued metric space.

AMS(2010): 47H10, 54H25.

1. Introduction

Fixed point theory plays a very crucial role in the development of nonlinear analysis. The
Banach [2] fixed point theorem for contraction mapping has been generalized and extended in
many directions. This famous theorem can be stated as follows.

Theorem 1.1([2]) Let (X, d) be a complete metric space and T be a mapping of X into itself
satisfying:
d(T x, T y) d(x, y), x, y X, (1.1)

where is a constant in [0, 1). Then T has a fixed point p X.

The Banach contraction principle with rational expressions have been expanded and some
fixed point and common fixed point theorems have been obtained in [4, 5].
Recently, Azam et al. [1] introduced the concept of complex valued metric space and estab-
lished some fixed point results for mappings satisfying a rational inequality. Complex-valued
metric space is useful in many branches of mathematics, including algebraic geometry, number
theory, applied mathematics; as well as in physics, including hydrodynamics, thermodynamics,
mechanical engineering and electrical engineering, for more details, see, [7, 8].
In this paper, we establish common fixed point results for generalized contraction involving
rational expression in the framework of complex valued metric spaces.

1 Received June 24, 2016, Accepted February 12, 2017.

54 G. S. Saluja

2. Preliminaries

Let C be the set of complex numbers and z1 , z2 C. Define a partial order - on C as follows:

z1 - z2 if and only if Re(z1 ) Re(z2 ), Im(z1 ) Im(z2 ). It follows that z1 - z2 if one of

the following conditions is satisfied:

(i) Re(z1 ) = Re(z2 ), Im(z1 ) < Im(z2 );

(ii) Re(z1 ) < Re(z2 ), Im(z1 ) = Im(z2 );
(iii) Re(z1 ) < Re(z2 ), Im(z1 ) < Im(z2 );
(iv) Re(z1 ) = Re(z2 ), Im(z1 ) = Im(z2 ).
In particular, we will write z1  z2 if z1 6= z2 and one of (i), (ii), or (iii) is satisfied and we
will write z1 z2 if only (iii) is satisfied. Note that

0 . z1  z2 |z1 | < |z2 |,

z1 - z2 , z2 z3 z1 z3 .

The following definition was introduced by Azam et al. in 2011 (see, [1]).

Definition 2.1([1]) Let X be a nonempty set. Suppose that the mapping d : X X C

satisfies:

(C1 ) 0 - d(x, y) for all x, y X with x 6= y and d(x, y) = 0 x = y;

(C2 ) d(x, y) = d(y, x) for all x, y X;
(C3 ) d(x, y) - d(x, z) + d(z, y) for all x, y, z X.
Then d is called a complex valued metric on X and (X, d) is called a complex valued metric
space.

Example 2.2 Let X = C, where C is the set of complex numbers. Define a mapping d : XX
C by d(z1 , z2 ) = eit |z1 z2 | where z1 = (x1 , y1 ), z2 = (x2 , y2 ) and t [0, 2 ]. Then (X, d) is a
complex valued metric space.

Example 2.3([1]) Let X = C, where C is the set of complex numbers. Define a mapping
d : X X C by d(z1 , z2 ) = e3i |z1 z2 | where z1 = (x1 , y1 ) and z2 = (x2 , y2 ). Then (X, d) is
a complex valued metric space.

Example 2.4 Let X = C. Define a mapping d : X X C by d(z1 , z2 ) = eia |z1 z2 | where

z1 = (x1 , y1 ), z2 = (x2 , y2 ) and a is any real constant. Then (X, d) is a complex valued metric
space.

Definition 2.5 (i) A point x X is called an interior point of a subset G X whenever

there exists 0 r C such that

B(x, r) = {y X : d(x, y) r} G.
 Fixed Point Results Under Generalized Contraction Involving Rational Expression in Complex Valued Metric Spaces 55

(ii) A point x X is called a limit of G whenever for every 0 r C such that

 
B(x, r) G {x} 6= .

(iii) The set G X is called open whenever each element of G is an interior point of G.
A subset H X is called closed whenever each limit point of H belongs to H.

The family F := {B(x, r) : x X, 0 r} is a sub-basis for a Hausdorff topology on X.

Definition 2.6([1]) Let (X, d) be a complex valued metric space. Let {xn } be a sequence in X
and x X. Then

(i) {xn } is called convergent, if for every c C, with 0 c there exists n0 N such that
for all n > n0 , d(xn , x) c. Also, {xn } converges to x (written as, xn x or limn xn = x)
and x is the limit of {xn }.
(ii) {xn } is called a Cauchy sequence in X, if for every c C, with 0 c there exists
n0 N such that for all n > n0 , d(xn , xn+m ) c. If every Cauchy sequence converges in X,
then X is called a complete complex valued metric space.

Definition 2.7([6]) Two families of self-mappings {Ti }m n

i=1 and {Si }i=1 are said to be pairwise
commuting if

(i) Ti Tj = Tj Ti , i, j {1, 2, , m};

(ii) Sk Sl = Sl Sk , k, l {1, 2, , n};
(iii) Ti Sk = Sk Ti , i {1, 2, . . . , m} and k {1, 2, , n}.

Lemma 2.8([1]) Let (X, d) be a complex valued metric space and let {xn } be a sequence in X.
Then {xn } converges to x if and only if limn |d(xn , x)| = 0.

Lemma 2.9([1]) Let (X, d) be a complex valued metric space and let {xn } be a sequence in X.
Then {xn } is a Cauchy sequence if and only if limn |d(xn , xn+m )| = 0.

3. Main Results

In this section we shall prove some common fixed point results under generalized contraction
involving rational expression in the framework of complex valued metric spaces.

Theorem 3.1 Let (X, d) be a complete complex valued metric space. Suppose that the mappings
S, T : X X satisfy:
h d(x, Sx)d(x, T y) + d(y, T y)d(y, Sx) i
d(Sx, T y) - d(x, y) +
d(x, T y) + d(y, Sx)
+ d(x, Sx) + d(y, T y)
+ [d(x, T y) + d(y, Sx)] (3.1)

for all x, y X, where , , , , are nonnegative reals with + + + + 2 < 1. Then S

56 G. S. Saluja

and T have a unique common fixed point in X.

Proof Let x0 be an arbitrary point in X and define

x2k+1 = Sx2k , x2k+2 = T x2k+1 , k = 0, 1, 2, .

Then from (3.1), we have

d(x2k+1 , x2k+2 ) = d(Sx2k , T x2k+1 )

- d(x2k , x2k+1 )
h d(x , Sx )d(x , T x
2k+1 ) + d(x2k+1 , T x2k+1 )d(x2k+1 , Sx2k )
i
2k 2k 2k
+
d(x2k , T x2k+1 ) + d(x2k+1 , Sx2k )
+ d(x2k , Sx2k ) + d(x2k+1 , T x2k+1 )
+ [d(x2k , T x2k+1 ) + d(x2k+1 , Sx2k )]
= d(x2k , x2k+1 )
h d(x , x
2k+1 )d(x2k , x2k+2 ) + d(x2k+1 , x2k+2 )d(x2k+1 , x2k+1 )
i
2k
+
d(x2k , x2k+2 ) + d(x2k+1 , x2k+1 )
+ d(x2k , x2k+1 ) + d(x2k+1 , x2k+2 )
+ [d(x2k , x2k+2 ) + d(x2k+1 , x2k+1 )]
- ( + + )d(x2k , x2k+1 ) + d(x2k+1 , x2k+2 )
+[d(x2k , x2k+1 ) + d(x2k+1 , x2k+2 )]
= ( + + + )d(x2k , x2k+1 ) + ( + )d(x2k+1 , x2k+2 ). (3.2)

This implies that

 + + + 
d(x2k+1 , x2k+2 ) - d(x2k , x2k+1 ). (3.3)
1

Similarly, we have

d(x2k+2 , x2k+3 ) = d(Sx2k+1 , T x2k+2 )

- d(x2k+1 , x2k+2 )
h d(x
2k+1 , Sx2k+1 )d(x2k+1 , T x2k+2 ) + d(x2k+2 , T x2k+2 )d(x2k+2 , Sx2k+1 )
i
+
d(x2k+1 , T x2k+2 ) + d(x2k+2 , Sx2k+1 )
+ d(x2k+1 , Sx2k+1 ) + d(x2k+2 , T x2k+2 )
+ [d(x2k+1 , T x2k+2 ) + d(x2k+2 , Sx2k+1 )]
= d(x2k+1 , x2k+2 )
h d(x
2k+1 , x2k+2 )d(x2k+1 , x2k+3 ) + d(x2k+2 , x2k+3 )d(x2k+2 , x2k+2 )
i
+
d(x2k+1 , x2k+3 ) + d(x2k+2 , x2k+2 )
+ d(x2k+1 , x2k+2 ) + d(x2k+2 , x2k+3 )
+ [d(x2k+1 , x2k+3 ) + d(x2k+2 , x2k+2 )]
- ( + + )d(x2k+1 , x2k+2 ) + d(x2k+2 , x2k+3 )
+[d(x2k+1 , x2k+2 ) + d(x2k+2 , x2k+3 )]
= ( + + + )d(x2k+1 , x2k+2 ) + ( + )d(x2k+2 , x2k+3 ). (3.4)
 Fixed Point Results Under Generalized Contraction Involving Rational Expression in Complex Valued Metric Spaces 57

This implies that

 + + + 
d(x2k+2 , x2k+3 ) - d(x2k+1 , x2k+2 ). (3.5)
1

Putting
 + + + 
h= .
1
As + + + + 2 < 1, it follows that 0 < h < 1, we have

d(xn+1 , xn+2 ) - h d(xn , xn+1 ) - - hn+1 d(x0 , x1 ). (3.6)

Let m, n 1 and m > n, we have

d(xn , xm ) - d(xn , xn+1 ) + d(xn+1 , xn+2 ) + d(xn+2 , xn+3 )

+ + d(xn+m1 , xm )
- [hn + hn+1 + hn+2 + + hn+m1 ]d(x1 , x0 )
h hn i
- d(x1 , x0 )
1h

and so
h hn i
|d(xn , xm )| |d(x1 , x0 )| 0 as m, n .
1h

This implies that {xn } is a Cauchy sequence. Since X is complete, there exists w X such
that xn w as n . It follows that w = Sw, otherwise d(w, Sw) = z > 0 and we would
then have

z - d(w, x2n+2 ) + d(x2n+2 , Sw) - d(w, x2n+2 ) + d(Sw, T x2n+1 )

- d(w, x2n+2 ) + d(w, x2n+1 )
h d(w, Sw)d(w, T x
2n+1 ) + d(x2n+1 , T x2n+1 )d(x2n+1 , Sw)
i
+
d(w, T x2n+1 ) + d(x2n+1 , Sw)
+ d(w, Sw) + d(x2n+1 , T x2n+1 ) + [d(w, T x2n+1 ) + d(x2n+1 , Sw)]
= d(w, x2n+2 ) + d(w, x2n+1 )
h d(w, Sw)d(w, x
2n+2 ) + d(x2n+1 , x2n+2 )d(x2n+1 , Sw)
i
+
d(w, x2n+2 ) + d(x2n+1 , Sw)
+ d(w, Sw) + d(x2n+1 , x2n+2 ) + [d(w, x2n+2 ) + d(x2n+1 , Sw)].

This implies that

|z| |d(w, x2n+2 )| + |d(w, x2n+1 )|

h |z||d(w, x
2n+2 )| + |d(x2n+1 , x2n+2 )||d(x2n+1 , Sw)|
i
+
|d(w, x2n+2 )| + |d(x2n+1 , Sw)|
+ |z| + |d(x2n+1 , x2n+2 )| + [|d(w, x2n+2 )| + |d(x2n+1 , Sw)|].
58 G. S. Saluja

Letting n , it follows that

|z| ( + )|z| ( + + + + 2)|z| < |z|

which is a contradiction and so |z| = 0, that is, w = Sw.

In an exactly the same way, we can prove that w = T w. Hence Sw = T w = w. This shows
that w is a common fixed point of S and T .

We now show that S and T have a unique common fixed point. For this, assume that w
is another common fixed point of S and T , that is, Sw = T w = w such that w 6= w . Then

d(w, w ) = d(Sw, T w )
h d(w, Sw)d(w, T w ) + d(w , T w )d(w , Sw) i
- d(w, w ) +
d(w, T w ) + d(w , Sw)
+ d(w, Sw) + d(w , T w ) + [d(w, T w ) + d(w , Sw)]

h d(w, w)d(w, w ) + d(w , w )d(w , w) i
= d(w, w ) +
d(w, w ) + d(w , w)
+ d(w, w) + d(w , w ) + [d(w, w ) + d(w , w)]

= ( + 2)(.w, w )

So that |d(w, w )| ( + 2)(.w, w ) < |d(w, w )|, since 0 < ( + 2) < 1, which is a
contradiction and hence d(w, w ) = 0. Thus w = w . This shows that S and T have a unique
common fixed point in X. This completes the proof. 2
Putting S = T in Theorem 3.1, we have the following result.

Corollary 3.2 Let (X, d) be a complete complex valued metric space. Suppose that the mapping
T : X X satisfies:
h d(x, T x)d(x, T y) + d(y, T y)d(y, T x) i
d(T x, T y) - d(x, y) +
d(x, T y) + d(y, T x)
+ d(x, T x) + d(y, T y) + [d(x, T y) + d(y, T x)] (3.7)

for all x, y X, where , , , , are nonnegative reals with + + + + 2 < 1. Then T

has a unique fixed point in X.

Corollary 3.3 Let (X, d) be a complete complex valued metric space. Suppose that the mapping
T : X X satisfies (for fixed n):
h d(x, T n x)d(x, T n y) + d(y, T n y)d(y, T n x) i
d(T n x, T n y) - d(x, y) +
d(x, T n y) + d(y, T n x)
+ d(x, T x) + d(y, T y) + [d(x, T n y) + d(y, T n x)]
n n
(3.8)

for all x, y X, where , , , , are nonnegative reals with + + + + 2 < 1. Then T

has a unique fixed point in X.
 Fixed Point Results Under Generalized Contraction Involving Rational Expression in Complex Valued Metric Spaces 59

Proof By Corollary 3.2, there exists q X such that T n q = q. Then

d(T q, q) = d(T T n q, T n q) = d(T n T q, T nq)

- d(T q, q)
h d(T q, T n T q)d(T q, T nq) + d(q, T n q)d(q, T n T q) i
+
d(T q, T n q) + d(q, T n T q)
+ d(T q, T T q) + d(q, T n q) + [d(T q, T n q) + d(q, T n T q)]
n

= d(T q, q)
h d(T q, T T nq)d(T q, T n q) + d(q, T n q)d(q, T T n q) i
+
d(T q, T n q) + d(q, T T n q)
+ d(T q, T T q) + d(q, T n q) + [d(T q, T n q) + d(q, T T nq)]
n

= d(T q, q)
h d(T q, T q)d(T q, q) + d(q, q)d(q, T q) i
+
d(T q, q) + d(q, T q)
+ d(T q, T q) + d(q, q) + [d(T q, q) + d(q, T q)]
= ( + 2) d(T q, q).

So that |d(T q, q)| ( + 2) |d(T q, q)| < |d(T q, q)|, since 0 < ( + 2) < 1, which is a
contradiction and hence d(T q, q) = 0. Thus T q = q. This shows that T has a unique fixed
point in X. This completes the proof. 2
As an application of Theorem 3.1, we prove the following theorem for two finite families of
mappings.

Theorem 3.4 If {Ti }m n

i=1 and {Si }i=1 are two finite pairwise commuting finite families of self-
mappings defined on a complete complex valued metric space (X, d) such that S and T (with
T = T1 T2 Tm and S = S1 S2 Sn ) satisfy the condition (3.1), then the component maps of
the two families {Ti }m n
i=1 and {Si }i=1 have a unique common fixed point.

Proof In view of Theorem 3.1 one can conclude that T and S have a unique common fixed
point g, that is, T (g) = S(g) = g. Now we are required to show that g is a common fixed
point of all the components maps of both the families. In view of pairwise commutativity of
the families {Ti }m n
i=1 and {Si }i=1 , (for every 1 k m) we can write

Tk (g) = Tk S(g) = STk (g) and Tk (g) = Tk T (g) = T Tk (g)

which show that Tk (g) (for every k) is also a common fixed point of T and S. By using the
uniqueness of common fixed point, we can write Tk (g) = g (for every k) which shows that g is
a common fixed point of the family {Ti }m
i=1 . Using the same arguments as above, one can also
show that (for every 1 k n) Sk (g) = g. This completes the proof. 2
By taking T1 = T2 = = Tm = G and S1 = S2 = = Sn = F , in Theorem 3.4, we
derive the following result involving iterates of mappings.
60 G. S. Saluja

Corollary 3.5 If F and G are two commuting self-mappings defined on a complete complex
valued metric space (X, d) satisfying the condition
h d(x, F n x)d(x, Gm y) + d(y, Gm y)d(y, F n x) i
d(F n x, Gm y) - d(x, y) +
d(x, Gm y) + d(y, F n x)
+ d(x, F x) + d(y, G y) + [d(x, Gm y) + d(y, F n x)]
n m
(3.9)

for all x, y X, where , , , , are nonnegative reals with + + + + 2 < 1. Then

F and G have a unique common fixed point in X.

By setting m = n and F = G = T in Corollary 3.5, we deduce the following result.

Corollary 3.6 Let (X, d) be a complete complex valued metric space and let the mapping
T : X X satisfies (for fixed n)
h d(x, T n x)d(x, T n y) + d(y, T n y)d(y, T n x) i
d(T n x, T n y) - d(x, y) +
d(x, T n y) + d(y, T n x)
+ d(x, T x) + d(y, T y) + [d(x, T n y) + d(y, T n x)]
n n
(3.10)

for all x, y X, where , , , , are nonnegative reals with + + + + 2 < 1. Then T

has a unique fixed point in X.

Proof By Corollary 3.2, we obtain p X such that T n p = p. The rest of the proof is same
as that of Corollary 3.3. This completes the proof. 2

By taking = h and = = = = 0 in Corollary 3.3, we draw following corollary

which can be viewed as an extension of Bryant (see, [4]) theorem to complex valued metric
space.

Corollary 3.7 Let (X, d) be a complete complex valued metric space. Suppose that the mapping
T : X X satisfying the condition

d(T n x, T n y) - h d(x, y)

for all x, y X and h [0, 1) is a constant. Then T has a unique fixed point in X.

The following example demonstrates the superiority of Bryant (see, [3]) theorem over Ba-
nach contraction theorem.

Example 3.8 Let X = C, where C is the set of complex numbers. Define a mapping d : CC
C by d(z1 , z2 ) = |x1 x2 | + i|y1 y2 | where z1 = x1 + iy1 , z2 = x2 + iy2 . Then (C, d) is a
 Fixed Point Results Under Generalized Contraction Involving Rational Expression in Complex Valued Metric Spaces 61

complex valued metric space. Define T : C C as

0, if x, y Q,

i, if x, y Qc ,


T (x + iy) =


1, if x Qc , y Q,

1 + i, if x Q, y Qc .

Now for x = 1 and y = 0, we get

2

1 1
d(T ( ), T (0)) = d(1, 0) - d( , 0) = .
2 2 2

Thus 2 which is a contradiction that 0 < 1. However, we notice that T 2 (z) = 0,
so that
0 = d(T 2 (z1 ), T 2 (z2 )) - d(z1 , z2 ),

which shows that T 2 satisfies the requirement of Bryant theorem and z = 0 is a unique fixed
point of T .

Finally, we conclude this paper with an illustrative example which satisfied all the condi-
tions of Corollary 3.2.

Example 3.9 Let X = {0, 12 , 2} and partial order - is defined as x - y iff x y. Let the
complex valued metric d be given as

d(x, y) = |x y| 2ei 4 = |x y|(1 + i) for x, y X.

Let T : X X be defined as follows:

1 1
T (0) = 0, T ( ) = 0, T (2) = .
2 2

Case 1. Take x = 12 , y = 0, T (0) = 0 and T ( 21 ) = 0 in Corollary 3.2, then we have

1 + i
d(T x, T y) = 0 ( + + + ).
2

This implies that = = = 0 and = = 12 or = = = 19 and = = 1

6 satisfied
all the conditions of Corollary 3.2 and of course 0 is the unique fixed point of T .

Case 2. Take x = 2, y = 21 , T (2) = 1

2 and T ( 21 ) = 0 in Corollary 3.2, then we have

1+i  3(1 + i)   3(1 + i)   3(1 + i) 

d(T x, T y) = . + . + .
2 2 2 2
1+i
+. + .2(1 + i).
2
1
This implies that = = = = = 13 satisfied all the conditions of Corollary 3.2 and of
62 G. S. Saluja

course 0 is the unique fixed point of T .

1
Case 3. Take x = 2, y = 0, T (2) = 2 and T (0) = 0 in Corollary 3.2, then we have

1+i  3(1 + i)   3(1 + i) 

d(T x, T y) = .2(1 + i) + . + .
2 2 2
5(1 + i)
+. .
2
1
This implies that = = = = 14 and = 0 satisfied all the conditions of Corollary 3.2
and of course 0 is the unique fixed point of T .

4. Conclusion

In this paper, we establish common fixed point theorems using generalized contraction involving
rational expression in the setting of complex-valued metric spaces and give an example in
support of our result. Our results extend and generalize several results from the current existing
literature.

References

[1] A. Azam, B. Fisher and M. Khan, Common fixed point theorems in complex valued metric
spaces, Numer. Funct. Anal. Optim., 3(3) (2011), 243-253.
[2] S. Banach, Surles operation dans les ensembles abstraits et leur application aux equation
integrals, Fund. Math., 3 (1922), 133-181.
[3] V. W. Bryant, A remark on a fixed point theorem for iterated mappings, Amer. Math.
Monthly, 75 (1968), 399-400.
[4] B. Fisher, Common fixed points and constant mapping satisfying rational inequality, Math.
Sem. Notes (Univ. Kobe), (1978).
[5] B. Fisher and M. S. Khan, Fixed points, common fixed points and constant mappings,
Studia Sci. Math. Hungar., 11 (1978), 467-470.
[6] M. Imdad, J. Ali and M. Tanveer, Coincedence and common fixed point theorem for
nonlinear contractions in Menger PM spaces, Chaos Solitones Fractals, 42 (2009), 3121-
3129.
[7] W. Sintunavarat and P. Kumam, Generalized common fixed point theorem in complex
valued metric spaces with applications, J. Ineql. Appl., doi:10.1186/1029-242X-2012-84.
[8] R. K. Verma and H. K. Pathak, Common fixed point theorem using property (E.A) in
complex valued metric spaces, Thai J. Math., 11(2) (2013), 347-355.
Math.Combin.Book Ser. Vol.1(2017), 63-70

A Study on Cayley Graphs over Dihedral Groups

A.Riyas and K.Geetha

(Department of Mathematics, T.K.M College of Engineering ,Kollam-691005, India)

E-mail: [email protected], [email protected]

Abstract: Let G be the dihedral group Dn and Cay(G, S) is the Cayley graph of G with
respect to S , and let CG (x) is the centralizer of an element x in G and x
is the orbit of x in
G. In this paper, we prove that if G act on G by conjugation, the vertex induced subgraph
with vertex set CG (x) of the Cayley graph Cay(G, x
) is either Hamiltonian or it contain
Hamiltonian decompositions. But if n is prime, it is always Hamiltonian.

Key Words: Smarandache-Cayley graph, Cayley graph, dihedral group, Hamiltonian

cycle, complete graph.

AMS(2010): 05C25.

1. Introduction

Let (G; ) be a finite group. A Smarandache-Cayley graph of G respect to a pair {S, T } of

non-empty subsets S G, T G \ S is the graph with vertex set G and edge set consisting of
pairs (x, y) such that s x = t y, where s S and t T . Particularly, let T = {1G }. Then such
a Smarandache-Cayley graph is the usual Cayley graph Cay(G, S), whose vertex set is G and
edges are the pairs (x, y) such that s x = y for some s S and x 6= y. Arthur Cayley (1878)
introduced the Cayley graphs of groups and it has received much attention in the literature.
Brian Alspach et al. (2010) proved that every connected Cayley graphs of valency at least three
on a generalized dihedral group, whose order is divisible by four is Hamilton-connected, unless
it is bipartite. Recently Adrian Pastine and Daniel Jaume (2012) proved that given a dihedral
group DH and a generating subset S, if S H 6= , then the Cayley digraph Cay(DH , S) is
Hamiltonian. In this paper, we denote a group (G; ) by G for convenience.

2. Main Results

In this section we deals with some basic definitions and terminologies of group theory and graph
theory which are needed in sequel. For details see Fraleigh (2003), Gallian (2009) and Diestel
(2010).

1 Received June 8, 2016, Accepted February 15, 2017.

64 A.Riyas and K.Geetha

Definition 2.1 Let G be a group. The orbit of an element x under G is usually denoted as x

and is defined as x = {gx/g G}.

Definition 2.2 Let x be a fixed element in a group G . The centralizer of an element x in G ,

CG (x) is the set of all element in G that commute with x. In symbols, CG (x) = {g G/gx =
xg}.

Definition 2.3 A group G act on G by conjugation means gx = gxg 1 for all x G.

Definition 2.4 An element x in a group G is called an involution if x2 = e.

Definition 2.5 The nth dihedral group Dn is the group of symmetries of the regular n-gon and
Dn Sn , where Sn is the symmetric group of n letters for n 3 with |Dn | = 2n.
The structure of Dn is {g, g 2 , g 3 , , g n , y, yg, yg 2, yg 3 , , yg n1 }, where g denote rota-
tion by 2n and y be any one of reflections (reflections along perpendicular bisector of sides
or along diagonal flips). Dn can be represented as G1 G2 where G1 =< g > and G2 =
{y, yg, yg 2, yg 3 , , yg n1 }. We say g and y are generators of Dn , and the equations g n =
y 2 = e , the identity and yg = g n1 y are relations for these generators. Generally all reflec-
tions are involutions and rotations may or may not. If n is odd, e is the only involution in G1
and G2 consist of reflections along perpendicular bisector of sides only. Except for e , generally
n
G1 and G2 never commute and G2 is non-abelian , but if n is even, g 2 is the only involution
in G1 which commute G2 .

Definition 2.6 A subgraph (U, F ) of a graph (V, E) is said to be vertex induced subgraph if F
consist of all the edges of (V, E) joining pairs of vertices of U .

Definition 2.7 A Hamiltonian path is a path in (V, E) which goes through all the vertices in
(V, E) exactly ones. A Hamiltonian cycle is a closed Hamiltonian path. A graph (V, E) is said
to be Hamiltonian, if it contains a Hamiltonian cycle.

Theorem 2.8 Let G be the dihedral group Dp , p is prime and G act on G by conjugation.
Then for every element x G1 with x 6= e , the induced subgraph with vertex set CG (x) of the
Cayley graph Cay(G, x
) is Hamiltonian.

Proof Given G = Dp , so G = {g, g 2 , g 3 , , g p , y, yg, yg 2, , yg p1 }. Since x G1 , we

have CG (x) = {x, x2 , x3 , , xp }. Let u CG (x). Then ux = xu for x G. x is the orbit of
x G with x2 6= e and G act on G by conjugation, we have x = {x, xp1 }, since CG (x) is
abelian and yx = xn1 y. We can choose an element s x such that s = (ux)x(ux)1 . Now
su = (ux)x(ux)1 u = (ux)x(x1 u1 )u = (ux)(xx1 )(u1 u) = ((ux)e)e = (ux), then there is
an edge from u to ux. Again s(ux) = (ux)x(ux)1 ux = ((ux)x) = ux2 , then there is an edge
from ux to ux2 and consequently a path from u to ux2 . Continuing in this way, we get a finite
path u ux ux2 ux3 uxp = ue = u in the induced subgraph with vertex set
CG (x) of the Cayley graph Cay(G, x), which is Hamiltonian. In particular for u = e , we get a
Hamiltonian cycle e x x2 x3 xp = e. 2
Definition 2.9 A graph (V, E) is said to be complete if for eah pair of arbitrary vertices in
 A Study on Cayley Graphs over Dihedral Groups 65

(V, E) can be joined by an edge. A complete graph of n vertices is denoted as Kn .

Theorem 2.10 Let G be the dihedral group D2n+1 and G act on G by conjugation. Then
for every element x G2 , the induced subgraph with vertex set CG (x) of the Cayley graph
Cay(G, x) is K2 .

Proof Given G = D2n+1 , so we have G = {g, g 2 , g 3 , , g 2n+1 , y, yg, yg 2, , yg 2n }. Since

x G2 , which is non-abelian, we have CG (x) = {x, e}. Let u CG (x). Then ux = xu for
x G. Since x is the orbit of x G2 and G act on G by conjugation, we have x x. We
can choose the element s = x x such that s = (ux)x(ux)1 . Now su = (ux)x(ux)1 u =
(ux)x(x1 u1 )u = (ux)(xx1 )(u1 u) = ((ux)e) = (ux), then there is an edge from u to ux.
Again s(ux) = (ux)x(ux)1 ux = ((ux)x)e = ux2 , so there exist an edge from ux to ux2 and
consequently a path from u to ux2 . Since x2 = e, we get a Hamiltonian cycle u ux ux2 =
ue = u in the induced subgraph with vertex set CG (x) of the Cayley graph Cay(G, x), which
is K2 . 2
Corollary 2.11 Let G be the dihedral group Dp , where p is prime and G act on G by conjuga-
tion. Then for every element x G2 , the induced subgraph with vertex set CG (x) of the Cayley
graph Cay(G, x) is Hamiltonian.

Theorem 2.12 Let G be the dihedral group Dp ,where p is prime and G act on G by conjugation.
Then for x G with x 6= e, the induced subgraph with vertex set CG (x) of the Cayley graph
Cay(G, x) is Hamiltonian.

Proof Since |G| = 2p, we have an element x G such that either xp = e or x2 = e. So

there exists a Hamiltonian cycle u ux ux2 ux3 uxp = u by Theorem 2.8 or a
Hamiltonian cycle u ux ux2 = ue = u by Theorem 2.10 in the induced subgraph with
vertex set CG (x) of the Cayley graph Cay(G, x). 2
Definition 2.13 A graph (V, E) is called bipartite if V = V1 V2 with V1 V2 = , and every
edge of (V, E) is of the form {a, b} with a V1 and b V2 .

Theorem 2.14 Let G be the dihedral group Dn and G act on G by conjugation. Then for every
element x G1 with x 6= e and CG (x) = G , the induced subgraph with vertex set CG (x) of the
Cayley graph Cay(G, x) is a bipartite graph on n vertices.

Proof Given G = Dn , so we have G = {g, g 2 , g 3 , , g n , y, yg, yg 2, yg 3 , , yg n1 }. Since

n
x G1 with CG (x) = G, we have either x = e or x = g 2 . But x 6= e. Let u CG (x).
Then ux = xu for all u G. x is the orbit of x G1 and G act on G by conjugation
,we have x = {x}. We can choose the element s = x x such that s = (ux)x(ux)1 . Now
1 1 1 1 1
su = (ux)x(ux) u = (ux)x(x u )u = (ux)(xx )(u u) = ((ux)e) = (ux), then there is
an edge from u to ux. Again s(ux) = (ux)x(ux)1 ux = ((ux)x)e = ux2 , then there is an edge
n
from ux to ux2 and consequently a path from u to ux2 . Since x = g 2 , we have x2 = e. Thus
we get a complete graph u ux u in the induced subgraph with vertex set CG (x) of the
Cayley graph Cay(G, x).
66 A.Riyas and K.Geetha

Let us consider the following cases.

Case 1. If u = g i , i = 1, 2, 3, , n, we get n2 distinct complete graph of two vertices with one

n n n
end vertex in {g, g 2 , g 3 , , g 2 } and other in {g 2 +1 , g 2 +2 , , g n } as shown below.
n n n n n n
g g 2 +1 g, g 2 g 2 +2 g 2 , , g 2 g n g 2 , g 2 +1 g g 2 +1 , ,
n
g g 2 gn.
n

Case 2. If u = yg i ,i = 1, 2, 3, , n, we get another n2 distinct complete graph of two vertices

n n n
with one end vertex in {yg, yg 2, yg 3 , , yg 2 } and other in {yg 2 +1 , yg 2 +2 , , yg n } as shown
below.
n n n n n n
yg yg 2 +1 yg, yg 2 yg 2 +2 yg 2 , , yg 2 y yg 2 , yg 2 +1 yg yg 2 +1 , ,
n
yg yg 2 yg n .
n

Thus the induced subgraph with vertex set CG (x) of the Cayley graph Cay(G, x) is a
bipartite graph on n vertices. 2
Remark 2.15 By Theorem 2.14 ,the graphs in case.2 have been completely characterized. If
n = pq with p and q are distinct primes, the induced subgraph with vertex set CG (x) of the
Caylay graph Cay(G, x) has n2 distinct complete graphs on two vertices with one end vertex in
{y, yg 2 , , yg n2 } and other in {yg, yg 3, , yg n1 }.
If n 6= pq, we get n2 distinct complete graph on two vertices. Out of which n4 graphs have one
n n n
end vertex in {y, yg 2, yg 4 , , yg 2 2 } and other in {yg 2 , yg 2 +2 , , yg n2 } and the remaining
n n n n
3 5 2 1 } and others in {yg 2 +1 , yg 2 +3 , ,
4 graphs have one end vertex in {yg, yg , yg , , yg
yg n1 }.

Corollary 2.16 Let G be the dihedral group Dn , where n is even and G act on G by conjugation.
n
Then for the element x = g 2 G, the induced subgraph with vertex set CG (x) of the Cayley
graph Cay(G, x) is a bipartite graph on n vertices.

Theorem 2.17 Let G be the dihedral group D4n and G act on G by conjugation. Then for
every involuted element x G with CG (x) 6= G, the induced subgraph with vertex set CG (x) of
the Cayley graph Cay(G, x) is Hamiltonian.

Proof Let G = D4n . So we have G = {g, g 2 , , g 4n , y, yg, yg 2, , yg 4n1 }. Since

n
x G with x2 = e and CG (x) 6= G, we have x 6= e and x 6= g 2 . Thus x G2 and

CG (x) = {x, e, xg 2n , g 2n }. We decompose G1 as G1 G1 where G1 = {g 2 , g 4 , , g 4n } and G1 =

{g, g 3 , , g 4n1 }. Similarly G2 an be decomposed as G2 G2 , where G2 = {y, yg 2, yg 4 , ,

yg 4n2 } and G2 = {yg, yg 3, , yg 4n1 }. Since x G2 , we have either x G2 or x G2 .

If x G2 , it implies that xg 2n G2 . From the composition table and also from the relation

yg = g 4n1 y, we get G1 G2 (G1 )1 = G1 G2 (G1 )1 = G2 G2 (G2 )1 = G2 G2 (G2 )1 = G2 . Thus

= G2 . Similarly if x G2 ,implies that xg 2n G2 . From the composition table it follows that
x

G1 G2 (G1 )1 = G1 G2 (G1 )1 = G2 G2 (G2 )1 = G2 G2 (G2 )1 = G2 and hence x = G2 .
Let u CG (x). Then ux = xu for x G. We can choose two involutions s1 and s2 in
such that s1 = (ux)x(ux)1 and s2 = (uxg 2n )xg 2n (uxg 2n )1 . Now s1 u = (ux)x(ux)1 u =
x
(ux)x(x1 u1 )u = (ux)(xx1 )(u1 u) = ((ux)e) = (ux), then there is an edge from u to ux.
 A Study on Cayley Graphs over Dihedral Groups 67

Again s2 (ux) = (uxg 2n )xg 2n (uxg 2n )1 (ux) = (uxg 2n )xg 2n (g 2n )1 x1 u1 ux = (uxg 2n )x =

(ux)g 2n x = u(xg 2n )x = u(g 2n x)x = (ug 2n )x2 = ug 2n , then there is an edge from ux to ug 2n and
consequently a path from u to ug 2n . Again s1 (ug 2n ) = (ux)x(ux)1 (ug 2n ) = (ux)xx1 u1 ug 2n =
uxg 2n , then there is an edge from ug 2n to uxg 2n and consequently a path from u to uxg 2n .
Again s2 (uxg 2n ) = (uxg 2n )xg 2n (uxg 2n )1 (uxg 2n ) = (uxg 2n )xg 2n = (uxg 2n )g 2n x = uxg 4n x =
ux2 = ue = u. Thus we get a Hamiltonian cycle u ux ug 2n uxg 2n u in the induced
subgraph with vertex set CG (x) of the Cayley graph Cay(G, x). In particular for u = e, we get
a Hamiltonian cycle e x g 2n xg 2n e. 2
Corollary 2.18 Let G be the dihedral group D4n and G act on G by conjugation. Then for
every x G2 , the induced subgraph with vertex set CG (x) of the Cayley graph Cay(G, x) is
Hamiltonian.

Theorem 2.19 Let G be the dihedral group D4n and G act on G by conjugation. Then for
every involuted element x G with CG (x) 6= G, the induced subgraph with vertex set CG (x) of
the Cayley graph Cay(G, x g 2n ) is K4 .

Proof Since x D4n with x2 = e and CG (x) 6= G by Theorem 2.17, we get a Hamiltonian
cycle u ux ug 2n uxg 2n u in the induced subgraph with vertex set CG (x) of the
Cayley graph Cay(G, x). To prove that this graph is K4 , it is enough to show that there
exist edges from u ug 2n and ux uxg 2n . We can choose s = g 2n as ug 2n u1 . Now su =
(ug 2n u1 )u = ug 2n , then there is an edge from u to ug 2n . Similarly we get an edge from ux to
uxg 2n , since s(ux) = (ug 2n u1 )ux = ug 2n x = uxg 2n . 2
Corollary 2.20 Let G be the dihedral group D4n and G act on G by conjugation. Then for
every x G2 , the induced subgraph with vertex set CG (x) of the Cayley graph Cay(G, x g 2n )
is K4 .

Theorem 2.21 Let G be the dihedral group D4n+2 and G act on G by conjugation. Then for
every involuted element x G with CG (x) 6= G, the induced subgraph with vertex set CG (x) of
the Cayley graph Cay(G, x) is a bipartite graph on four vertices.

Proof Given G = D4n+2 , so we have G = {g, g 2 , g 3 , , g 4n+2 , y, yg, yg 2, , yg 4n+1 }.

Since x G with x2 = e and CG (x) 6= G, clearly x G2 and hence CG (x) = {x, e, g 2n+1 , xg 2n+1 }.

Since x G2 , either x G2 or x G2 , where G2 = {y, yg 2 , , yg 4n } and G2 =
{yg, yg 3, , yg 4n+1 }. If x G2 , then xg 2n+1 G2 . Since x is the orbit of an element

x in G2 and G act on G by conjugation, we get x = G2 . Similarly if x G2 , we have

xg 2n+1 G2 and x = G2 . Thus there exist exactly one involution in x CG (x). We can choose
that s x such that s = (ux)x(ux)1 .
Let u CG (x). Then ux = xu for x G. Now su = (ux)x(ux)1 u = (ux)x(x1 u1 )u =
(ux)(xx1 )(u1 u) = ((ux)e) = (ux), then there is an edge from u to ux. Again s(ux) =
(ux)x(ux)1 ux = ((ux)x)e = ux2 = ue = u. Thus we get a Hamiltonian cycle u ux u in
the induced subgraph with vertex set CG (x) of the Cayley graph Cay(G, x). Since |CG (x)| = 4 ,
there exist an element other than u and ux in CG (x). Since ug 2n+1 commute with all reflections,
68 A.Riyas and K.Geetha

we have ug 2n+1 CG (x). Again s(ug 2n+1 ) = (ux)x(ux)1 (ug 2n+1 ) = (ux)(xx1 )(u1 u)g 2n+1 =
uxg 2n+1 and on the other hand s(uxg 2n+1 ) = (ux)x(ux)1 (uxg 2n+1 ) = (ux)(xx1 )(u1 u)xg 2n+1 =
ux2 g 2n+1 = ug 2n+1 . Thus we get another cycle ug 2n+1 uxg 2n+1 ug 2n+1 in the induced
subgraph with vertex set CG (x) of the Cayley graph Cay(G, x). Thus the induced subgraph
with vertex set CG (x) of the Cayley graph Cay(G, x) is a bipartite graph on four vertices. 2

Corollary 2.22 Let G be the dihedral group D4n+2 and G act on G by conjugation. Then for
every x G2 , the induced subgraph with vertex set CG (x) of the Cayley graph Cay(G, x) is a
bipartite graph on four vertices.

Theorem 2.23 Let G be the dihedral group D4n+2 and G act on G by conjugation. Then for
every involuted element x G with CG (x) 6= G, the induced subgraph with vertex set CG (x) of
the Cayley graph Cay(G, x g 2n+1 ) is Hamiltonian.

Proof Since G = D4n+2 and G act on G by conjugation, by Theorem 2.21, for every
x G with CG (x) 6= G, the induced subgraph with vertex set CG (x) of the Cayley graph
Cay(G, x) is a bipartite graph on 4 vertices with one cycle u ux u and another cycle
ug 2n+1 uxg 2n+1 ug 2n+1 . If we add an element g 2n+1 in x, then we get an edge from u to
ug 2n+1 and ux to uxg 2n+1 , since g 2n+1 u = ug 2n+1 and g 2n+1 (ux) = (ux)g 2n+1 . Thus we get
a Hamiltonian cycle ug 2n+1 u ux uxg 2n+1 ug 2n+1 in the induced subgraph with
vertex set CG (x) of the Cayley graph Cay(G, x g 2n+1 ). 2
Corollary 2.24 Let G be the dihedral group D4n+2 and G act on G by conjugation. Then for
every x G2 , the induced subgraph with vertex set CG (x) of the Cayley graph Cay(G, xg 2n+1 )
is Hamiltonian.

Theorem 2.25 Let G be the dihedral group Dn , n is even and G act on G by conjugation. Then
n
for every x G2 , the induced subgraph with vertex set CG (x) of the Cayley graph Cay(G, xg 2 )
is Hamiltonian.

Proof Suppose G = D4n and G act on G by conjugation. Then by Corollary 2.20, for every
x G2 , the induced subgraph with vertex set CG (x) of the Cayley graph Cay(G, x g 2n ) is
K4 . Also we have if G = D4n+2 and G act on G by conjugation, by Corollary 2.24, for every
x G2 , the induced subgraph with vertex set CG (x) of the Cayley graph Cay(G, x g 2n+1 ) is
Hamiltonian. Thus if G = Dn , n is even ,we get for every x G2 , the induced subgraph with
n
vertex set CG (x) of the Cayley graph Cay(G, x g 2 ) is Hamiltonian. 2
Theorem 2.26 Let G be the dihedral group Dn and G act on G by conjugation. Then for
x G with x = g m , the induced subgraph with vertex set CG (x) of the Cayley graph Cay(G, x)
is Hamiltonian if gcd(m, n) = 1.

Proof Given G = Dn , so we have G = {g, g 2 , g 3 , , g n , y, yg, yg 2 , , yg n1 }. Since x G

with x = g m and gcd(m, n) = 1, we get CG (x) = {x, x2 , x3 , , xn } and x = {x, xn1 } . As
in the proof Theorem 2.8, we get a Hamiltonian cycle u ux ux ux3 uxn =
2

ue = u in the induced subgraph with vertex set CG (x) of the Cayley graph Cay(G, x). 2
 A Study on Cayley Graphs over Dihedral Groups 69

Theorem 2.27 Let G be the dihedral group Dn and G act on G by conjugation. Then for every
element x G with x = g m with CG (x) 6= G, induced subgraph with vertex set CG (x) of the
) has d Hamiltonian decompositions on nd vertices if gcd(m, n) = d.
Cayley graph Cay(G, x

Proof Given G = Dn , so we have G = {g, g 2 , g 3 , , g n , y, yg, yg 2, , yg n1 }. Since

n
x G with x = g m and CG (x) 6= G, we have m 6= n2 and n. Thus x G1 other than g 2 and
g n and hence CG (x) = {x, x2 , x3 , , xm1 , xm , xm+1 , , xn }. Let u CG (x). Then ux = xu
for x G. x is the orbit of x G and G act on G by conjugation, we have x = {x, xn1 }.
Choose an element s = x x such that s = (ux)x(ux)1 . Now su = (ux)x(ux)1 u = (ux),
then there is an edge from u to ux. Again s(ux) = (ux)x(ux)1 ux = ux2 , then there is an
edge from ux to ux2 and consequently a path from u to ux2 . Continuing in this way, we
n
get a cycle u ux ux2 ux d = u in the induced subgraph with vertex set
CG (x) of the Cayley graph Cay(G, x). In particular, for u = g i ,i = 1, 2, , n, we get d
mn
Hamiltonian decompositions on nd vertices as g g 1+m g 1+2m g 1+ d = g,
mn mn
g 2 g 2+m g 2+2m g 2+ d = g 2 , , g d g d+m g d+2m g d+ d =
g d , g d+1 g d+1+m g d+1+2m g d+1 , . . . , g n g m g n of which the
decompositions when u = g i and u = g i+d are same. 2
Theorem 2.28 Let G be the dihedral group Dn and G act on G by conjugation. Then for every
element x G1 with x = g m and x 6= e , induced subgraph with vertex set CG (x) of the Cayley
graph Cay(G, x) has d Hamiltonian decompositions on nd vertices if gcd(m, n) = d.

Proof Given G = Dn and G act on G by conjugation.Then by Theorem 2.27, for every

element x G with x = g m with CG (x) 6= G, induced subgraph with vertex set CG (x) of the
Cayley graph Cay(G, x) has d Hamiltonian decompositions on nd vertices if gcd(m, n) = d. Also
we have ,by Theorem 2.14, for every x G1 with x 6= e and CG (x) = G , the induced subgraph
with vertex set CG (x) of the Cayley graph Cay(G, x) is a bipartite graph on n vertices. Thus
if G = Dn and G act on G by conjugation, for every element x G1 with x = g m and x 6= e,
induced subgraph with vertex set CG (x) of the Cayley graph Cay(G, x) has d Hamiltonian
decompositions on nd vertices if gcd(m, n) = d. 2

References

[1] Adrian Pastine and Danial Jaume, On Hamiltonian circuits in Cayley graph over general-
ized dihedral groups, Union Mathematica Argentina, 53(2012),79-87.
[2] Brian Alspach, C.C.Chen and Matthew Deans, Hamiltonian paths in Cayley graph on
generalized dihedral groups, ARS Mathematica Contemporanea, 3(2010), 29-47.
[3] Diestel.R, Graph Theory, Graduate Texts in Mathematics, Newyork, 1997.
[4] John B.Fraleigh, A First Course in Abstrat Algebra, Seventh Edition, Pearson Educa-
tion,Inc, 2003.
[5] Joseph A.Gallian, Contemporary Abstract Algebra, Narosa Publications, Seventh Edition,
2004.
[6] Riyas A and K.Geetha, A study on Hamiltonian property of Cayley graphs over non-abelian
70 A.Riyas and K.Geetha

groups, International Journal of Mathematical Combinatorics, Vol.3, (2016), 142-146.

[7] Riyas A and K.Geetha, A study on Cayley graphs of non-abelian groups, International
Journal of Mathematical Combinatorics, Vol.4, (2016), 79-87.
Math.Combin.Book Ser. Vol.1(2017), 71-77

On the Second Order Mannheim Partner Curve in E 3

Seyda Klcoglu
(Department of Mathematics, Baskent University, Turkey)

S
uleyman Senyurt
(Department of Mathematics, Ordu University, Turkey)

E-mail: [email protected]

Abstract: In this study first we worked on the Mannheim curve pair {, 1 } and Mannheim
curve pair {1 , 2 } We called 2 as the second order Mannheim partner curve of the
Mannheim curve . We examined the Frenet apparatus of second order Mannheim partner
curve in terms of, Frenet apparatus of Mannheim curve , with the offset property of second
order Mannheim partner 2 . Further we examined third order Mannheim partner 3 where
{2 , 3 } are Mannheim curve pair.

Key Words: Mannheim curve, Frenet apparatus, second Mannheim curve, modified Dar-
boux vector.

AMS(2010): 53A40, 53B30.

1. Introduction

Mannheim curve was firstly defined by A. Mannheim in 1878. A curve is called a Mannheim

curve if and only if 2 is a nonzero constant, is the curvature and is the torsion.
+ 2
Mannheim curve was redefined in [6], if the principal normal vector N of first curve and binormal
vector B1 of second curve are linearly dependent, then first curve is called Mannheim curve,
and the second curve is called Mannheim partner curve. As a result they called these new
curves as Mannheim partner curves. For more detail see in [6]. Frenet-Serret apparatus of the
curve : I E 3 are {T, N, B, , }. For any unit speed curve , the Darboux and modified
Darboux vectors are, respectively ([2],[4])

D(s) = (s)T (s) + (s)B (s) , (1.1)

D(s) = (s)T (s) + B (s) . (1.2)

In [7] Mannheim curves are studied and Mannheim partner curve of can be represented

(s1 ) = 1 (s1 ) + (s1 )B1 (s1 ) (1.3)

1 Received June 15, 2016, Accepted February 16, 2017.

72 S
eyda Klco
glu and S
uleyman S
enyurt

for some function , since N and B1 are linearly dependent, equation can be rewritten as

1 (s) = (s) (s)N (s) , (1.4)

where
(s)
(s) =
2
2 . (1.5)
(s) + (s)
Frenet-Serret apparatus of Mannheim partner curve 1 are {T1 , N1 , B1 , 1 , 1 }. The rela-
tionship and 1 Frenet vectors are as follows

T1 = cos T sin B
N1 = sin T + cos B (1.6)
B1 = N.

where (T, T1 ) = cos . The first curvature and the second curvature (torsion) are

d
1 = = , 1 = . (1.7)
ds1 cos

We use dot to denote the derivative with respect to the arc length parameter of the curve
. Also
ds 1 1
= = , (1.8)
ds1 cos sin
for more detail see in [7], or we can write

ds 1
= . (1.9)
ds1 1 +

2. Second Order Mannheim Partner and Frenet Apparatus

Definition 2.1 Let { , 1 } and {1 , 2 } be the Mannheim pairs of and 1 respectively.

We called as 2 is a second order Mannheim partner of the curve . which has the following
parametrization ,
2 = + 1 sin T N + 1 cos B, (2.1)

where
1 = (s) N (s) and 2 = 1 (s) 1 N1 (s) . (2.2)

Theorem 2.1 The Frenet vectors of second order Mannheim partner 2 of a Mannheim curve
 On the Second Order Mannheim Partner Curve in E 3 73

, based on the Frenet apparatus of Mannheim curve are

T2 = cos 1 cos T sin 1 N cos 1 sin B


N2 = sin 1 cos T + cos 1 N sin 1 sin B (2.3)


B2 = sin T + cos B.

Proof Let 2 be second order Mannheim partner of a Mannheim curve . Also 2 be the
Mannheim partner of Mannheim partner 1 . The Frenet vector fields T1 , N1 , B1 and T2 , N2 , B2
which are belong to the curves 1 and 2 , respectively. It is easy to say that Frenet vectors of
second order Mannheim partner 2 , based on the Frenet vectors of Mannheim curve 1 are

T2 = cos 1 T1 sin 1 B1


N2 = sin 1 T1 + cos 1 B1


B2 = N1

where (T1 , T2 ) = 1 . By substituting T1 , N1 , B1 we have the equalities in terms of the curve

.
T2 = cos 1 (cos T sin B) sin 1 N
N2 = sin 1 (cos T sin B) + cos 1 N
B2 = sin T + cos B

This completes the proof. Also the following product give us the same equalities;

T2 cos 1 0 sin 1 cos 0 sin T
2

N2 = sin 1 0 cos 1 sin 0 cos N .

B2 0 1 0 0 1 0 B

Theorem 2.2 Let 2 be second order Mannheim partner of a Mannheim curve . The curvature
and torsion of the second order Mannheim partner 2 are

1 1
2 = , 2 = . (2.4)
cos cos 1 cos 1

Proof Since 1 = and 1 = , we have the curvature as in the following way
cos

d1 1
2 = = .
ds2 cos cos 1

Also as in the following way we have the torsion

1
2 = = .
1 1 cos 1

We use mark to denote the derivative with respect to the parameter of the curve . Due
74 S
eyda Klco
glu and S
uleyman S
enyurt

to this theorem we also get

ds 1
= . (2.5)
ds2 cos cos 1
2
Theorem 2.3 The modified Darboux vector of Mannheim partner 1 of a Mannheim curve ,
is
2
1 (s) = cos T + N cos sin B
D (2.6)

Proof Similarly from the equation (1.2)

1 (s) = 1 T1 (s) + B1 (s) .

D (2.7)
1

Substituting the equation (2.7) into equation (1.6)and (1.7), the proof is complete. 2
Theorem 2.4 The modified Darboux vector of second order Mannheim partner 2 of a Mannheim
curve , is

cos2 1 cos
 
2 cos 1 sin 1
D = + sin T N
1 1
cos2 1 sin
 
cos B. (2.8)
1

Proof Since
2 (s) = 2 T2 (s) + B2 (s) .
D (2.9)
2
Substituting the equation (2.9) into equation (2.3)and (2.4), the proof is complete. 2
Theorem 2.5 The offset property of second order Mannheim partner 2 can be given if and
only if the curvature and the torsion of satisfy the following equation

cos
1 = 2 , (2.10)
2 + (2 + 2 ) cos2

2
where 2 + 2 + 2 cos2 6= 0.

with the offset property 1 = 1 21 + 12 and


Proof Notice that 1 = cos ,1 =

1
21 + 12


=
1
1
1 =
cos 2 + cos2 2 + 2 2

cos2

cos
1 = 2 .
2 + (2 + 2 ) cos2
 On the Second Order Mannheim Partner Curve in E 3 75

This completes the proof. 2

Theorem 2.6 The second order Mannheim partner 2 is not a Mannheim partner curve .

Proof Since the definition of Mannheim partner curve,

hB2 (s) , N (s)i = h sin T + cos B, N i = 0,

hence N (s) and B2 (s) are linear independent. 2

Definition 2.2 Let { , 1 } , {1 , 2 } and {2 , 3 } be the Mannheim pairs of , 1 and 2

respectively. We called as 3 is a third order Mannheim partner of the curve , which has the
following parametrizations,

3 (s) = 2 (s) 2 N2 (s)

= + (1 sin + 2 sin 1 cos ) T ( 2 cos 1 ) N
+ (1 cos 2 sin 1 sin ) B, (2.11)

where
2 = + 1 sin T N + 1 cos B (2.12)

and
| + 1 + 2 |

is the distance between the arclengthed curves and 3 .

Theorem 2.7 The Frenet vectors of third order Mannheim partner 3 of a Mannheim curve
, based on the Frenet apparatus of Mannheim curve are



T3 = (cos 2 cos 1 cos sin 2 sin ) T cos 2 sin 1 N





(sin 2 cos + cos 2 cos 1 sin ) B






N3 = (sin 2 cos 1 cos + cos 2 sin ) T sin 2 sin 1 N (2.13)


+ (cos 2 cos sin 2 cos 1 sin ) B












B = sin cos T + cos N sin sin B
3 1 1 1

where (T2 , T3 ) = cos 2 .

76 S
eyda Klco
glu and S
uleyman S
enyurt

Proof Since

T3 cos 2 0 sin 2 cos 1 0 sin 1

N3
=
sin 2 0 cos 2 sin 1 0
cos 1

B3 0 1 0 0 1 0

cos 0 sin T

sin 0 cos N ,

0 1 0 B

we have the proof. 2

Corollary 2.1 The product of Frenet vector fields of third order Mannheim partner 3 and
Mannheim curve , has the following matrix form

cos 2 cos 1 cos cos 2 sin 1 sin 2 cos

sin sin cos 2 cos 1 sin
2



[V3 ]T [V] =

sin 2 cos 1 cos sin 2 sin 1 cos 2 cos
(2.14)

+ cos 2 sin
sin 2 cos 1 sin




sin 1 cos cos 1 sin 1 sin

where [V3 ] = [T3 , N3 , B3 ] and [V] = [T, N, B].

Corollary 2.2 Let 3 be third order Mannheim partner of a Mannheim curve . The curvature
and torsion of the third order Mannheim partner 3 are

2 1 1
3 = , 3 = . (2.15)
cos cos 1 cos 2 cos 1 2

Proof We can write

d2 2
3 = =
ds3 cos cos 1 cos 2
and
2 1
3 = = 1
2 2 cos 1 2
or also since
2
cos cos 1 cos 2 =
3
and
1
cos cos 1 =
2
. 2
 On the Second Order Mannheim Partner Curve in E 3 77

References

[1] Boyer C., A History of Mathematics, Wiley, New York 1968.

[2] Gray A., Modern Differential Geometry of Curves and Surfaces with Mathematica, 2nd ed.
Boca Raton, FL: CRC Press, p. 205, 1997.
on
[3] Hacsalihoglu H.H., Differential Geometry(in Turkish) , In u University Publications,
Malatya, vol.1, 1994.
[4] Izumiya S., Takeuchi N., Special curves and ruled surfaces, Beitr age zur Algebra und
Geometrie Contributions to Algebra and Geometry, Vol.44, No.1(2003), 203-212.
[5] Lipschutz M.M., Differential Geometry, Schaums Outlines.
[6] Liu H., Wang F., Mannheim partner curves in 3-space, Journal of Geometry, Vol.88, No.1-
2(2008), 120-126(7).
[7] Orbay K., Kasap E., On Mannheim partner curves E 3 , International Journal of Physical
Sciences, Vol.4, No.5(2009), 261-264.
[8] McCleary J., Geometry From a Differentiable View Point, Vassar Collage, Cambridge
University Press 1994.
[9] Schief W.K., On the integrability of Bertrand curves and Razzaboni surfaces, Journal of
Geometry and Physics, Vol.45, No.1-2(2003), 130-150.
Math.Combin.Book Ser. Vol.1(2017), 78-87

The Change of Special Finsler Spaces

H. S. Shukla, O. P. Pandey and Khageshwar Manda

(Department of Mathematics & Statistics, DDU Gorakhpur University, Gorakhpur, India)

E-mail: [email protected], [email protected], [email protected]

Abstract: We have considered the change of Finsler metric L given by L = f (L, ),

where f is any positively homogeneous function of degree one in L and . We have ob-
tained the change of Creducible Finsler spaces, S3like Finsler spaces and T tensor.
Particular case when bi in is concurrent vector field has been studied.

Key Words: change, Finsler metric, T tensor, Creducible, S3like Finsler spaces.

AMS(2010): 53B20, 53B28, 53B40, 53B18, 53C60.

1. Introduction

Let F n = (M n , L) be an ndimensional Finsler space on the differentiable manifold M n ,

equipped with the fundamental function L(x, y). B. N. Prasad and Bindu Kumari [1] and C.
Shibata [2] considered the change of Finsler metric given by

L (x, y) = f (L, ), (1.1)

where f is positively homogeneous function of degree one in L and and given by (x, y) =
bi (x) y i is a one-form on M n . The Finsler space (M n , L ) obtained from F n by the change
(1.1) will be denoted by F n . The Homogeneity of f in (1.1) gives

Lf1 + f2 = f, (1.2)

where the subscripts 1 and 2 denote the partial derivatives with respect to L and respec-
tively.
Differentiating (1.2) with respect to L and respectively, we get

Lf11 + f12 = 0 and Lf12 + f22 = 0.

Hence, we have
f11 f12 f22
2
= = 2,
L L

1 Received July 19, 2016, Accepted February 18, 2017.

 The Change of Special Finsler Spaces 79

which gives
f11 = 2 , f22 = L2 , f12 = L,

where the Weierstrass function is positively homogeneous function of degree 3 in L and .

Therefore
L1 + 2 + 3 = 0. (1.3)

Again 2 is positively homogeneous of degree 4 in L and , so

L21 + 22 + 42 = 0. (1.4)

Throughout the paper we frequently use above equations (1.2) to (1.4) without quoting
them. The concept of concurrent vector field has been given by Matsumoto and K. Eguchi [6]
and S. Tachibana [7], which is defined as follows:

The vector field bi is said to be a concurrent vector field if

(i) bi|j = gij , (ii) bi |j = 0, (1.5)

where small and long solidus denote the h and vcovariant derivatives respectively.
It has been proved by by Matsumoto that bi and its contravariant components bi are
functions of coordinates alone. Therefore from (1.5)(ii), we have

Cijk bi = 0.

2. Fundamental Quantities of F n

To find the relation between fundamental quantities of F n and F n , we use the following results

i = bi , i L = li , j li = L1 hij , (2.1)

where i stands for and hij are components of angular metric tensor of F n given by
y i
hij = gij li lj = Lj j L.
The successive differentiation of (1.1) with respect to y i and y j gives:

li = f1 li + f2 bi , (2.2)

f f1
hij = hij + f L2 mi mj , (2.3)
L

where mi = bi L li . The quantities corresponding to F n will be denoted by putting star on
the top of those quantities.
From (2.2) and (2.3) we get the following relations between metric tensors of F n and F n

f f1 p
gij = gij li lj + (f L2 + f22 )bi bj + p(li bj + lj bi ), (2.4)
L L
80 H. S. Shukla, O. P. Pandey and Khageshwar Manda

where p = (f1 f2 f L).

The contravariant components of the metric tensor of F n will be dertived from (2.4) as
follows:

pL3 f L4 i j pL2
 
L ij
g ij = g + 3 2
f i j
2 l l b b 2 (li bj + lj bi ), (2.5)
f f1 f f1 t L f f1 t f f1 t

where we put bi = g ij bj , li = g ij lj , b2 = g ij bi bj and

2
t = f1 + L3 , = b2 . (2.6)
L2
Putting q = 3f2 + f2 , we find that
f
(a) i f = li + f2 mi ,
L
(b) i f1 = Lmi ,
(c) i f2 = L2 mi , (2.7)
3
(d) i = li + 2 mi ,
L
(e) i b2 = 2C..i ,
2
(f ) i = 2C..i 2 mi
L
and

(a) i p = Lqmi ,
(b) i t = 2L3 C..i + (L3 2 3L)mi , (2.8)
3q
(c) i q = li + (4f2 2 + 3 2 L2 + f 22 )mi ,
L
where . denotes the contraction with bi , viz. C..i = Cjki bj bk .
Differentiating (2.4) with respect to y k , using (2.1) and (2.7), we get the following relation
between the Cartans Ctensors (Cijk
= 12 k gij

and Cijk = 12 k gij ):

f f1 p qL2
Cijk = Cijk + (hij mk + hjk mi + hki mj ) + mi mj mk . (2.9)
L 2L 2
It is to be noted that

mi li = 0, mi mi = = mi b i , hij lj = 0, hij mj = hij bj = mi , (2.10)

i
where mi = g ij mj = bi Ll .
i
To find Cjk = g ih Cjhk

we use (2.5), (2.9), (2.10), we get

i i p qL3
Cjk = Cjk + (hjk mi + hij mk + hik mj ) + mj mk mi
2f f1 2f f1
(2.11)
L pL 2pL + L4 q
C.jk ni 2 hjk ni mj mk n i ,
ft 2f f1 t 2f 2 f1 t

where ni = f L2 bi + pli .
 The Change of Special Finsler Spaces 81

We have the following relations corresponding to the vectors with components ni and mi :

Cijk mi = C.jk , Cijk ni = f L2 C.jk , mi mi = f L2 . (2.12)

3. The Change of Creducible Finsler Space

Let F n be a Creducible Finsler space. Then [5]

1
Chjk = (hhj Ck + hhk Cj + hjk Ch ), (3.1)
n+1

where Ck = Chjk g hj .
Using equation (3.1) in equation (2.9), we get

qL2
Chjk = (pk hhj + pj hhk + ph hjk ) + mh mj mk , (3.2)
2
where
f f1 p
pk = Ck + mk . (3.3)
L(n + 1) 2L
Using equation (2.3) in equation (3.2), we get

L
Chjk = (pk hhj + pj hhk + ph hjk ) + qh mj mk + qj mh mk + qk mj mh , (3.4)
f f1

where
qL2 L3
qh = mh ph . (3.5)
6 f1
Now suppose that the transformation (1.1) is such that (n+1)(f1 2 +3L 2 )mh = 6f1 Ch ,
then qh = 0. So equation (3.4) reduces to

L
Chjk = (pk hhj + pj hhk + ph hjk ) (3.6)
f f1

Ck L
which will give = pk , so that
n+1 f f1

1
Chjk = (C h + Cj hhk + Ch hjk ) (3.7)
n + 1 k hj

Hence F n is also a Creducible. Therefore we have the following result.

Theorem 3.1 Under the change of Finsler metric with the condition (n + 1)(f1 2 +
3L 2 )mh = 6f1 Ch , the Creducible Finsler space is transformed to a Creducible Finsler
space.

In the theorem (3.1) we have assumed that (n + 1)(f1 2 + 3L 2 )mh = 6f1 Ch . However
if this condition is not satisfied then a Creducible Finsler space may not be transformed to
82 H. S. Shukla, O. P. Pandey and Khageshwar Manda

a Creducible Finsler space. In the following we discuss under what condition a Creducible
Finsler space is transformed to a Creducible Finsler space by change of Finsler metric.

In both the spaces F n and F n are Creducible then from (3.1) and its corresponding
equation for F n we find, on using (2.9), that

f L2
[(Qh mj mk + Qj mh mk + Qk mj mh ) f1 (C..h hjk + C..j hhk
t  
p f f1 r
+ C..k hjh )] = (hjk mh + hhj mk + hhk mj ) (3.8)
2L L(n + 1)
 2 
qL 2
+ 3f L r mh mj mk ,
2

where Qh = tCh L3 C..h and r = (n 2)pt + f1 (3p + L3 q). Thus, we have the following
result.

Theorem 3.2 A Creducible Finsler space is transformed to a Creducible Finsler space by

a change of Finsler metric if and only if (3.8) holds.

The condition (3.8) of theorem (3.2) is too complicated to study any geometrical concept
of Finsler space. So we consider that our in change of Finsler metric is such that bi is a
concurrent vector field [6] so that C.i = 0, C..i = 0. Hence equation (3.8) reduces to
 
p f f1 r
f L2 (Ch mj mk + Cj mh mk + Ck mj mh ) = (hjk mh + hhj mk
2L 2L
 2 
qL
+ hhk mj ) + 3f 2 r mh mj mk .
2

Contracting this equation with g jk , we find

2f L3 Ch = {(n + 1)(p f f1 r) + (qL3 6f 2 Lr)} mh .

Hence we have the following result.

Theorem 3.3 If a Creducible Finsler space is transformed to a Creducible Finsler space

by a concurrent change of Finsler metric, then the vector Ch is along the direction of the
vector mh .

4. The Change of vCurvature Tensor

To find the vcurvature tensor of F n with respect to Cartans connection, we use the following:

h
Cij hhk = Cijk , hkj hik = hij , hij ni = f L2 mj . (4.1)
 The Change of Special Finsler Spaces 83

The vcurvature tensors Shijk of F n [4] is defined as

r r
Shijk = Chk Crij Chj Cikr . (4.2)

From (2.9), (2.10), (2.11), (2.12), (2.13) and (2.14) we get the following relation between
vcurvature tensors of F n and F n [1]:

f f1
Shijk = Shijk + dhj dik dhk dij + Ehk Eij Ehj Eik , (4.3)
L
where r
s pf1 2p qf1
dij = L C.ij hij + Lmi mj , (4.4)
t 2L2 ts 2 ts
p p qf1
Eij = hij Lmi mj (4.5)
2L2 f 2f1 f
and s = f f1 .
Now suppose that bi is a concurrent vector field and F n is an S3like Finsler space [4],
then C.ij = 0,
S
Shijk = 2 (hhk hij hhj hik ),
L
where S is any scalar function of x and y.
In view of these equations, we have from (4.3)

p2 f12 p2
 
f f1 S
Shijk= + (hhk hij hhj hik )
L3 4L4 ts 4L4 f
 
p(p qf1 ) pf1 (2p qf1 ) (4.6)
+ (hhj mi mk + hik mh mj
4L2 f f1 4Lts
hhk mi mj hij mh mk ).

Now suppose that the transformed Finsler space F n is also S3like. Then

S
Shijk = (h h hhj hik ). (4.7)
L2 hk ij
Now from (2.3), it follows that
 2
f f1
(hhk hij hhj hik ) = (hhk hij hhj hik )
L (4.8)
2
+ f f1 L(hhk mi mj + hij mh mk hhj mk mi hik mh mj ).

In view of (4.6), (4.7) and (4.8), we have

p2 f12 p2 S f12
 
f f1 S
+ (hhk hij hhj hik )
L3 4L4 ts 4L4 f L2
 
p(p qf1 ) pf1 (2p qf1 ) (4.9)
+ S f1 L (hhk mi mj
4L2 f f1 4Lts
+ hij mh mk hhj mi mk hik mh mj ) = 0.
84 H. S. Shukla, O. P. Pandey and Khageshwar Manda

Contracting (4.9) by g ij g hk , we get

p2 f12 p2 S f12
 
f f1 S
+ (n 1)(n 2)
L3 4L4 ts 4L4 f L2
  (4.10)
p(p qf1 ) pf1 (2p qf1 )
+2 S f 1 L = 0.
4L2 f f1 4Lts

Hence, we have the following result.

Theorem 4.1 If a S3like Finsler space is transformed to a S3like Finsler space under the
concurrent change, then equation (4.10) holds.

5. The Change of T Tensor

The T tensor of Finsler space F n is defined by [3]:

Thijk = LChij |k + lh Cijk + li Chjk + lj Chik + lk Chij , (5.1)

where
Chij r r r
Chijk |k = Crij Chk Chrj Cik Chir Cjk . (5.2)
y k
To find the T tensor of F n , first of all we find


Chij r r r
Chij ||k = Crij Chk Chrj Cik Chir Cjk ,
y k

where || denotes vcovariant derivative in F n . The derivatives of mi and hij with respect to
y k are given by
1
k mi = 2 hik li mk ,
L L (5.3)
1
k (hij ) = 2Cijk (li hjk + lj hki ).
L
From (2.7), (2.8), (2.9) and (5.3), we get

Chij f f1 Chij p
k
= k
+ (Chij mk + Cijk mh + Cjhk mi + Cihk mj )
y L y L
p p
(hij hhk + hhj hik + hih hjk ) + (hjk lh mi + hhk lj mi
2L3 2L2
+ hhk li mj + hik lh mj + hjk li mh + hik lj mh + hij lh mk + hhj li mk
q (5.4)
+ hih lj mk + hij lk mh + hjh lk mi + hhi lk mj )
(hij mh mk
2
+ hjh mi mk + hhi mj mk + hik mj mh + hjk mi mh + hhk mi mj )
qL
(li mj mh mk + lj mh mi mk + lh mi mj mk + lk mi mj mh )
2
L2
+ (4f2 2 + 3L2 2 + f 22 ) mi mj mh mk .
2
 The Change of Special Finsler Spaces 85

From equation (2.9), (2.10), (2.11) and (2.12), we have

r f f1 r p
Crij Chk = Crij Chk + (Chjk mi + Chik mj + Chij mk
L 2L
f1 p f f1 L 2
+ Cijk mh ) + (C.ij hhk + C.hk hij ) C.ij C.hk
2Lt t
p2 L2 (qf1 2p)
+ hij hhk + (C.ij mh mk + C.hk mi mj )
4f Lt 2t (5.5)
p(p + L3 q) p2
+ (hij mk mh + hhk mi mj ) + (hij mk mh
4Lf t 4Lf f1
+ hhk mi mj + hjk mi mk + hjk mi mh + hih mj mk + hik mj mh )
L2 {2pqt + (qf1 2p)(2p + L3 q)}
+ mi mj mh mk .
4f f1 t

From equation (5.4) and (5.5), we get

f f1 p
Chij ||k = Chij |k (Chij mk + Cijk mh + Chjk mi + Cihk mj )
L 2L
p(2f t + L2 p)

q
(h h
ij hk + h h
hj ik + h h
ih jk )
4f L3 t 2
2 3 2

f1 p + f1 L pq + 3p
+ (hij mk mh + hhk mi mj + hjh mi mk
4Lf f1 t
p 
+ hik mj mh + hhi mj mk + hjk mi mh ) lh (hjk mi
2L2
+ hij mk + hik mj ) + lj (hhk mi + hik mh + hih mk ) + li (hhk mj
(5.6)
qL
+ hjk mh + hhj mk ) + lk (hij mh + hjh mi + hhi mj ) (li mj mh mk
2
f1 p
+ l j mh mi mk + l h mi mj mk + l k mi mj mh ) (C.ij hhk + C.hj hik
2Lt
f f1 L 2
+ C.hk hij + C.ik hhj + C.hi hjk + C.jk hhi ) + (C.ij C.hk
t
2
L (qf1 2p)
+ C.hj C.ik + C.hi C.jk ) (C.ij mk mh + C.hk mi mj
2t
+ C.hj mi mk + C.ik mj mh + C.hi mj mk + C.jk mh mi )

L2

+ (4f2 2 + 3L2 2 + f 22 )
2
3L2 {2pqt + (qf1 2p)(2p + L3 q)}

mi mj mh mk .
4f f1 t

Using equations (2.2), (2.9) and (5.6), we get the following relation between T tensors of
86 H. S. Shukla, O. P. Pandey and Khageshwar Manda

Finsler spaces F n and F n :

f 2 f1 f (f1 f2 + f L)
Thijk = Thijk + (Chij mk + Cijk mh + Chjk mi
L2 2L
f 2 L 2 f1 f f1 p
+ Cihk mj ) + (C.ij C.hk + C.hj C.ik + C.hi C.jk )
t 2Lt
(C.ij hhk + C.hk hij + C.hj hik + C.ik hhj + C.hi hjk + C.jk hhi )
f L2 (qf1 2p)
(C.ij mk mh + C.hk mi mj + C.hj mi mk
2t
p(2f t + L2 p)
+ C.ik mj mh + C.hi mj mk + C.jk mh mi ) (hij hhk (5.7)
4L3 t
f1 p2 + f1 L3 pq + 3p2
 
f q pf2
+ hhj hik + hih hjk ) +
2 4Lf1t L
(hij mk mh + hhk mi mj + hjh mi mk + hik mj mh + hhi mj mk
 2
fL 4L2 f2 q
+ hjk mi mh ) + (4f2 2 + 3L2 2 + f 22 ) +
2 2
2 3

3L {2pqt + (qf1 2p)(2p + L q)}
mi mj mh mk .
4f1 t

If bi is a concurrent vector field in F n , then C.ij = 0. Therefore from (5.7), we have

f 2 f1 p(2f t + L2 p)
Thijk = 2
Thijk (hij hhk + hhj hik + hih hjk )
L 4L3 t
f q f1 p2 + f1 L3 pq + 3p2
 
pf2
+
2 4Lf1 t L
(hij mk mh + hhk mi mj + hjh mi mk + hik mj mh + hhi mj mk (5.8)
 2
fL 4L2 f2 q
+ hjk mi mh ) + (4f2 2 + 3L2 2 + f 22 ) +
2 2
2 3

3L {2pqt + (qf1 2p)(2p + L q)}
mi mj mh mk .
4f1 t

If bi is a concurrent vector field in F n , with vanishing T tensor then T tensor of F n is

given by
p(2f t + L2 p)
Thijk = (hij hhk + hhj hik + hih hjk )
4L3 t
f q f1 p2 + f1 L3 pq + 3p2 pf2
 
+
2 4Lf1 t L
(hij mk mh + hhk mi mj + hjh mi mk + hik mj mh + hhi mj mk (5.9)
 2
fL 4L2 f2 q
+ hjk mi mh ) + (4f2 2 + 3L2 2 + f 22 ) +
2 2
2 3

3L {2pqt + (qf1 2p)(2p + L q)}
mi mj mh mk .
4f1 t
 The Change of Special Finsler Spaces 87

References

[1] B. N. Prasad and Bindu Kumari, The change of Finsler metric and imbedding classes
of their tangent spaces, Tensor N. S., 74(2013), 48-59.
[2] C. Shibata, On invariant tensors of )change of Finsler metric, J. Math. Kyoto Univ.,
24(1984), 163-188.
[3] F. Ikeda, On the tensor Tijkl of Finsler spaces, Tensor N. S., 33(1979), 203-209.
[4] F. Ikeda, On S3like and S4like Finsler spaces with the T tensor of a special form,
Tensor N. S., 35 (1981), 345-351.
[5] M. Matsumoto, On Creducible Finsler spaces, Tensor N. S., 24(1972), 29-37.
[6] M. Matsumoto and K. Eguchi, Finsler space admitting a concurrent vector field, Tensor
N. S., 28(1974), 239-249.
[7] S. Tachibana, On Finsler spaces which admit a concurrent vector field, Tensor N. S.,
1(1950), 1-5.
Math.Combin.Book Ser. Vol.1(2017), 88-101

Peripheral Distance Energy of Graphs

Kishori P. Narayankar and Lokesh S. B.

(Department of Mathematics, Mangalore University, Mangalagangothri, Mangalore-574199, India)

E-mail: kishori [email protected], [email protected]

Abstract: The peripheral distance matrix of a graph G of order n with k peripheral vertices
is a square symmetric matrix of order k k, denoted as Dp -matrix of G and is defined as
Dp (G) = [dij ], where dij is the distance between two peripheral vertices vi and vj in G. The
peripheral distance energy of a graph G is the sum of the absolute values of the eigenvalues
of Dp -matrix of G. The sum of the distances between all pairs of peripheral vertices is a
peripheral Wiener index of a graph G. In this paper, we study some preliminary facts of
Dp -matrix of G and give some bounds for peripheral distance energy of a graph G. Specially
the bounds are presented for a graph of diameter less than 3. Bounds of peripheral distance
energy in terms of peripheral Wiener index are also obtained for graphs of diam(G) 2.

Key Words: Distance, peripheral Wiener index, peripheral distance matrix, peripheral
distance energy.

AMS(2010): 05C12, 05C50.

1. Introduction

Let G be a connected, nontrivial graph with vertex set V (G) and edge set E(G) and let
|V (G)| = n and |E(G)| = m. Let u and v be two vertices of a graph G. The distance d(u, v|G)
between the vertices u and v is the length of a shortest path connecting u and v. If u = v then
d(u, v|G) = 0. The eccentricity e(v) of a vertex v in a graph G is the distance between v and
a vertex farthest from v in G. The diameter diam(G) of G is the maximum eccentricity of G,
while the radius rad(G) is the smallest eccentricity of G. A vertex v with e(v) = diam(G) is
called a peripheral vertex of G. The set of peripheral vertices of G is called as periphery and is
denoted as P (G).
We claim that the adjacency matrix of a graph is the distance based matrix such that the
entries of adjacency matrix are 1 if the distance between two vertices is 1 and 0 otherwise.
The distance matrix of a graph G is defined as a square matrix D = D(G) = [dij ], where
dij is the distance between vi and vj in G. For the application and the background of the
distance matrix on the chemistry, one can refer to [1, 32].
Peripheral distance matrix or Dp -matrix, Dp of a graph G is defined as, Dp = Dp (G) =
1 Received June 10, 2016, Accepted February 24, 2017.
 Peripheral Distance Energy of Graphs 89

[dij ], where dij is the distance between two peripheral vertices vi and vj in G. The eigenvalues
1 , 2 , , k of the Dp -matrix are said to be Dp -eigenvalues of G denoted by Dp spec(G).
Since Dp -matrix of G is symmetric, all of its eigenvalues are real and can be arranged in a
non-increasing order as 1 2 k . Recalling the definition of peripheral distance
matrix, a graph G of order n with k peripheral vertices, the peripheral distance matrix of G is
a (k k) matrix, whose entries are as follows:

Dp (G) = [dij ] = [d(vi , vj )]; where vi , vj P (G).

The peripheral distance energy (Dp -energy (in short)) of a graph G is defined as the sum
of the absolute values of Dp - eigenvalues of Dp -matrix of G. i.e,

k
X
EDP (G) = |i |. (1)
i=1

The form of (1) is chosen so as to be fully analogous to the definition of graph energy [5, 6, 9].
n
X
E = E(G) = |i |, (2)
i=1

where 1 , 2 , , n are the ordinary eigenvalues [3], i.e the eigenvalues of the adjacency matrix
A(G). Observe that the graph energy E(G) in past a few years has been extensively studied
and surveyed in Mathematics and Chemistry [8, 11, 14, 18, 19, 20, 21, 22, 25, 26, 27, 29, 30,
31, 33]. Through out the paper |P (G)| = k with labellings v1 , v2 , , vk , where 2 k n.

The characteristic polynomial of Dp (G) is the det(I Dp (G)), it is referred to as a

characteristic polynomial of G and is denoted by (G; ) = c0 k + c1 k1 + c2 k2 + + ck .
The roots 1 , 2 , , k of the polynomial (G; ) are called the eigenvalues of Dp (G). The
eigenvalues of Dp (G) are said to be the peripheral distance eigenvalues (or Dp -eigenvalues (in
short)) of G. Since Dp (G) is a real symmetric matrix, the Dp -eigenvalues are real and can
be ordered in non-increasing order, 1 2 k . Then the Dp -spectrum of a graph
G is the set of eigenvalues of Dp (G), together with the multiplicities of Dp -eigenvalues of
Dp (G). If the Dp -eigenvalues of Dp (G) are 1 2 k and their multiplicities are
m(1 ), m(2 ), , m(k ), then we shall write

1 2 k
Dp spec(G) = .
m(1 ) m(2 ) m(k )

For example, let G be a graph as shown in Fig.1. Then

. v1 v2 v3

v 0 3 3
1
Dp (G) =


v2 3 0 2

v3 3 2 0
90 Kishori P. Narayankar and Lokesh S. B.

Clearly, the characteristic polynomial of G is (G; ) = 3 + 22 + 36, whose Dp - eigen-

values are 1 + 19, 1 19 and 2. Hence EDp -energy of G is 10.7178.
v1

v5

G: v6 v4

v2
v3

Fig.1 G is a graph of order n = 6 with k = 3 peripheral vertices.

This paper is organized as follows: In the forthcoming section some preliminary facts of
peripheral distance matrix Dp (G) of G are obtained. In section 3 bounds of peripheral distance
energy in terms peripheral Wiener index are deduced. In section 4 bounds for the peripheral
distance energy are established. In the last section the smallest peripheral distance energy of a
graph is obtained thereby posing an open problem for the maximum peripheral distance energy.

2. Preliminary Results

Lemma 2.1 Let G be a graph of order n with k peripheral vertices and let 1 , 2 , , k be its
peripheral distance eigenvalues. Then,

k
X
(1) i = 0;
i=1

k
2
X X
(2) i 2 = 2 (dij ) .
i=1 1i<jk

k
X k
X
Proof Since, i = trace[Dp (G)] but dii = 0 in Dp (G), therefore, i = 0.
i=1 i=1

For i = 1, 2, , k, the (i, i)th entry of [Dp (G)]2 is equal to

k k
2
X X
dij , dji = (dij )
i=1 j=1
 Peripheral Distance Energy of Graphs 91

since Dp (G) is symmetric. Therefore,

k
X
i 2 = trace[Dp (G)]2
i=1
k X
k
(dij )2 = 2 (dij )2
X X
=
i=1 j=1 i<j
k
2
X X
= i 2 = 2 (dij ) . (3)
i=1 i<j

2
Lemma 2.2 Suppose G is a graph of order n and size m with k peripheral vertices having the
diam(G) 2. Then,
k    
X n k
i 2 = 6 +2 6m.
i=1
2 2

Proof In the peripheral distance matrix Dp of G there are x = 2m 2{(n k)k +

(nk)(nk1)
2 } elements equal to unity, and y = k(k 1) x elements equal to two. Therefore,

k
X
i 2 = trace[Dp (G)]2
i=1
k X
k
2 2
X X
= (dij ) = 2 (dij )
i=1 j=1 i<j
k
X
= i 2 = (x).12 + (y).22
i=1
= (x).12 + (k(k 1) x).22
= 4.k(k 1) 3x
= 4.k(k 1) 3{2m + k(k 1) n(n 1)}
= k(k 1) + 3.n(n 1) 6m
k    
X n k
i 2 = 6 +2 6m.
i=1
2 2

3. Preliminary Results with Respect to Peripheral Wiener Index

Definition 3.1([4, 7]) The thorn graph of the graph G , with parameters t1 , t2 , , tn is obtained
by attaching ti new vertices of degree one to the vertex vi of the graph G; i = 1, 2, , n. The
thorn graph of the graph G will be denoted by G , or if the respective parameters need to be
specified, by G (t1 , t2 , , tn ).

Definition 3.2([7, 28]) The thorn graph of the graph G obtained by attaching t new vertices of
92 Kishori P. Narayankar and Lokesh S. B.

degree one to all the vertices vi of the graph G is denoted by G+t .

If we partition the vertex set V (G) of a graph into two sets, with peripheral vertices in
one set and non-peripheral vertices in other. Then the sum of the distances between all pairs
of peripheral vertices is the peripheral Wiener index of a graph G. More formally
X
P W I(G) = d(vi , vj |G), (4)
1i<jk

where G is an (n, m)-graph with k peripheral vertices and vi , vj P (G).

Theorem 3.3([17]) Suppose G is a graph of order n and size m with k peripheral vertices
having diam(G) 2. Then,
   
n k
P W I(G) = + m. (5)
2 2

Theorem 3.4 Suppose G is a graph of order n and size m with k peripheral vertices having
the diam(G) 2. Then, for G+t

tk      
X
2 n k
i = 4k + 14 + 18 14m t2 4kt.
i=1
2 2

Proof In the peripheral distance matrix Dp (G+t ) there are x1 = kt elements equal to
0, x2 = k(t2 t) elements equal to 2, x3 = t2 {2m + 2 k2 2 n2 } elements equal to 3 and

x4 = t2 {2 n2 2m} elements equal to 4. Therefore,

kt
X
i 2 = trace[Dp (G+t )]2
i=1
kt X
kt
(dij )2 = 2 (dij )2
X X
=
i=1 j=1 i<j
kt
X
= i 2 = (x1 ).02 + (x2 ).22 + (x3 )3 2 + (x4 ).42
i=1
          
2 22 k n 2 2 n
= {k(t t)}.2 + t 2m + 2 2 .3 + t 2 2m .42
2 2 2
          
 2
2 k n 2 n
= 4k(t t) + 9t 2m + 2 2 + 16t 2 2m
2 2 2
          
2 2 k n 2 n
= (4kt 4kt) + t 18m + 18 18 + t 32 32m
2 2 2
     
k 2 n 2 n 2
= 4kt2 4kt + 18mt2 + 18 t 18 t + 32 t 32mt2
2 2 2
kt      
X n k
i 2 = 4k + 14 + 18 14m t2 4kt. (6)
i=1
2 2
 Peripheral Distance Energy of Graphs 93

2
Corollary 3.5 Suppose G is a graph of order n and size m with k peripheral vertices having
the diam(G) 2. Then, for G+t

tk    
X
2 k
i = 4k + 4 + 14P W I(G) t2 4kt.
i=1
2

Proof The proof follows directly from Theorems 3.3 and 3.4. 2

Proposition 3.6 Suppose G(n, m) is a graph with k peripheral vertices and diam(G) 2.
Then,
k  
X k
i 2 = 6P W I(G) 4 ,
i=1
2

where PWI(G) is the peripheral Wiener index of G.

Proof From the Lemma 2.2 we have,

k    
X
2 n k
i = 6 +2 6m
i=1
2 2
       
n k k k
= 6 +2 6m + 4 4
2 2 2 2
      
n k k
= 6 + m 4
2 2 2
 
k
= 6 {P W I(G)} 4
2

from Theorem 3.3. 2

4. Bounds for the Peripheral Distance Energy

Theorem 4.1 Suppose G is a graph with k peripheral vertices. Then

s s
2 2
X X
2 (dij ) EDP (G) 2.k. (dij ) . (7)
i<j i<j

Proof We have from Cauchy-Schwarz inequality

k
!2 k
! k
!
X X X
ai b i ai 2 bi 2
i=1 i=1 i=1
94 Kishori P. Narayankar and Lokesh S. B.

Put ai = 1 and bi = |i | then

k
!2 k
! k
!
X X X
[EDP (G)]2 = |i | 1 i 2
i=1 i=1 i=1
k
!
X
= k i 2
i=1

X
= k 2 (dij )2
i<j

from Eq.(3) and s X

[EDP (G)] 2k (dij )2 . (8)
i<j

We have from the definition

k
!2 k
X X X
2
[EDP (G)] = |i | = i 2 + 2 |i ||j |
i=1 i=1 i<j
X X
= 2 (dij )2 + 2 |i ||j |,
i<j i<j

2
X X
[EDP (G)] = 2 (dij )2 + |i ||j |, (9)
i<j i6=j

2
X X
[EDP (G)] 2 (dij )2 = |i ||j |. (10)
i<j i6=j

Also, we know that

k
!2 k
2
X X X
2
[EDP (G)] = |i | |i | = 2 (dij )2
i=1 i=1 i<j
X
2
= [EDP (G)] 2 (dij )2
i<j
s
X
= [EDP (G)] 2 (dij )2 . (11)
i<j

Inequations (8) and (11) complete the proof. 2

Corollary 4.2 Suppose G is any graph with k peripheral vertices and diam(G) = d. Then,
p
k(k 1) EDP (G) d.k. k 1.

k


Proof Since d(vi , vj ) = dij 1, for i 6= j and totally 2 pairs of peripheral vertices in G
 Peripheral Distance Energy of Graphs 95

form lower bound of Corollary 4.2.

s  
k
s X
EDP (G) 2. (dij ) 2.[1]2
2

i<j
2
r
k(k 1)
= 2.1. ,
2
p
EDP (G) k(k 1). (12)

Also, d(vj , vj ) = dij d, for i 6= j and totally k2 pair of peripheral vertices in G form

upper bound of Corollary 4.2.

s  
k
s X
EDP (G) 2.k. (dij )2 2.k.[d]2
i<j
2
r
k(k 1)
= 2.k.[d]2
2

EDP (G) d.k. k 1. (13)

Inequations (12) and (13) complete the proof. 2

Theorem 4.3 Suppose G is any graph with k peripheral vertices. Then,

q P
(1) 2 i<j (dij )2 + k(k 1)2/k EDP (G);
q
(2) EDP (G) k2 i<j (dij )2 + (k 1)[2 i<j (dij )2 ( k2 i<j (dij )2 )2 ],
P P P

where is the absolute value of the determinant of the peripheral distance matrix DP (G).

Proof We know that, for non-negative numbers the arithmetic mean is not smaller than
the geometric mean.
1
k(k1) 1
k
! k(k1)
1 X Y Y
2(k1)
|i ||j | |i ||j | = |i |
k(k 1) i=1
i6=j i6=j

k
!2/k
= | det(Dp (G))|2/k = ()2/k
Y
= |i |
i=1

|i ||j | k(k 1). ()2/k

X
=
i6=j

2 2/k
X
= [EDP (G)] 2 (dij )2 k(k 1). ()
i<j

2 2/k
X
[EDP (G)] k(k 1). () +2 (dij )2 ,
i<j
96 Kishori P. Narayankar and Lokesh S. B.

s
2/k
X
[EDP (G)] k(k 1). () +2 (dij )2 . (14)
i<j

Therefore, the equation (14) proves lower bound.

To prove the upper bound we follow the ideas of Koolen and Moulton [18, 19], who obtained
an analogous upper bound for ordinary graph energy E(G). By applying the Cauchy-Schwartze
inequality to the two (k 1) vectors (1, 1, , 1) and (|1 |, |2 |, , |k |) we get.

k
!2 k
!
X X
|i | (k 1) 2i
i=2 i=2


2 X
EDp (G) 1 (k 1) 2 (dij )2 21
i<j
v
u
u X
EDp (G) 1 + t(k 1) 2 (dij )2 21
u
i<j

Define the function v

u
u X
f (x) = x + t(k 1) 2 (dij )2 x2
u
i<j

we set x = 1 and bear in mind that 1 1.

X s X
From Equation (3) we get x2 = 21 2 (dij )2 = x 2 (dij )2 .
i<j i<j
s X

Now f (x) = 0 implies, x = 2
k (dij )2 . Therefore f (x) is a decreasing function in the
i<j
interval s
2 X
sX
2
(dij ) x 2 (dij )2 .
k i<j i<j

and s
2 X 2X
(dij )2 (dij )2 1 .
k i<j k i<j

Hence
2 X
f (1 ) f (dij )2 .
k i<j

Hence the proof. 2

Theorem 4.4 Suppose G is a graph of order n and size m with k peripheral vertices having
the diam(G) 2. Then,
s     s      
n k n k
6 +2 6m EDP (G) k 6 +2 6m .
2 2 2 2
 Peripheral Distance Energy of Graphs 97

Proof From Theorem 4.1 we have

s s
(dij )2 EDP (G) (dij )2
X X
2 2.k.
i<j i<j

and next from Lemma 2.2,

k    
X 2
X n k
2 (dij ) = i 2 = 6 +2 6m.
i<j i=1
2 2

2
2
X
n k



By replacing the 2 (dij ) by 6 2 +2 2 6m. in Ineq.7 gives the proof.
i<j

Corollary 4.5 Suppose G is a graph with diam(G) 2. having k peripheral vertices. Then,
s   s   
k k
6P W I(G) 4 EDP (G) k. 6P W I(G) 4 ,
2 2

where P W I(G) is the peripheral Wiener index of a graph G.

Proof The proof follows from Theorem 4.1 and Proposition 3.6. 2
Theorem 4.6 Suppose G is any graph with k peripheral vertices and diam(G) 2. Then,
s   r
k 2/k
1 1
S+2 EDP (G) {S} + (k 1)[S ( {S})2 ],
2 k k

where is the absolute value of the determinant of the peripheral distance matrix DP (G) and
S = 6 n2 + 2 k2 6m.

Proof The proof follows from Theorem 4.3 and Lemma 3.4. 2
Corollary 4.7 Suppose G is any graph with k peripheral vertices and diam(G) 2. Then,
s   r
k 2/k
1 1
S+2 EDP (G) {S} + (k 1)[S ( {S})2 ],
2 k k

where is the absolute value of the determinant of the peripheral distance matrix DP (G),
S = 6P W I(G) 4 k2 and P W I(G) is the peripheral Wiener index of a graph G.

Proof The proof follows from Theorem 4.3 and Proposition 3.6. 2
Theorem 4.8 Suppose G is a graph of order n and size m with k peripheral vertices having
the diam(G) 2. Then,
p
T EDP (G+t ) kt {T},
98 Kishori P. Narayankar and Lokesh S. B.

n o
where T = 4k + 14 n2 + 18 k2 14m t2 4kt.

Proof The proof follows from Theorem 4.1 and Lemma 2.2. 2

Corollary 4.9 Suppose G is a graph of order n and size m with k peripheral vertices having
the diam(G) 2. Then,
p
T EDP (G+t ) kt {T},
n o
where T = 4k + 4 k2 + 14P W I(G) t2 4kt and P W I(G) is the peripheral Wiener index of

a graph G.

proof The proof follows from Theorem 4.1 and Corollary 3.5. 2

Theorem 4.10 Suppose G is any graph with k peripheral vertices and diam(G) 2. Then,
s   r
kt 2/kt +t 1 1
T+2 EDP (G ) {T} + (kt 1)[T ( {T})2 ],
2 kt kt

where nis the absolute value of theodeterminant of the peripheral distance matrix DP (G+t )
and T = 4k + 14 n2 + 18 k2 14m t2 4kt.

proof The proof follows from Theorem 4.3 and Lemma 3.4. 2

5. The Smallest Peripheral Distance Energy of a Graph

By studying the bounds for peripheral distance energy, there arise a common question that,
which n vertex graphs with k peripheral vertices have the smallest and greatest peripheral
distance energy. Among all n-vertex connected graphs with k peripheral vertices the complete
graph is the unique graph with the smallest peripheral distance energy.

Theorem 5.1 The complete graph Kn=k with k peripheral vertices is the graph with smallest
peripheral distance energy, which is equal to 2(k 1).

Proof Let G be a graph with k peripheral vertices and Kk be a complete graph on k

peripheral vertices. Let A be a peripheral distance matrix of Kk . B be a peripheral distance
matrix of G with the Dp -eigenvalues 1 , 2 , , k . Clearly A and B are non-negative matrices
and obviously 0 A B. Now, from the fact that if 0 A B then (A) (B). And for
the complete graph, (A) = n 1 and ED (Kk ) = 2(k 1) hence,

2(k 1) = 2(A) 2(B)

k
X
(B) + |i |.
i=2
 Peripheral Distance Energy of Graphs 99

By using Perron Frobenius theorem, it implies that (B) is a positive eigenvalues. Hence,

k
X
2(k 1) |i | = EDp (G).
i=1

But
2(k 1) = EDp (Kk ) EDp (G).

Hence, we conclude that the peripheral distance energy of a graph with k peripheral vertices
is greater than the peripheral distance energy of a complete graph on k vertices. This proves
that among k peripheral vertices graphs complete graph has the smallest peripheral distance
energy = 2(k 1). 2
Since, distance matrix D of a complete graph is equal to peripheral distance matrix Dp of
a complete graph, also distance energy ED of a complete graph is equal to peripheral distance
matrix EDp of a complete graph, therefore this also settles the conjecture posed by Ramane et al.
in [24]. However, in [2], the authors have given the direct reason for the proof of the conjecturer
in [24]. Since, we do not have a sufficient stuff to prove graph with greatest peripheral distance
energy, but the graph with k peripheral vertices such that all the peripheral vertices are at the
distance d (= diam(G)) from each other is certainly deserve to be seriously considered graph.
In this connection it looks plausible to pose an open problem:

Open Problem The graph G with k peripheral vertices such that all of its peripheral vertices
are at the same distance d (= diam(G)) from each other has maximum peripheral distance
energy.

Acknowledgement

The authors thank Department of Science and Technology(SERB), Government of India, for
supporting through SB/EMEQ-119/2013.

References

[1] A. T. Balaban, D. Ciubotariu, M. Medeleanu, Topological indices and real number vertex
invariants based on graph eigenvalues or eigenvectors, J. Chem. Inf. Comput. Sci., 31
(1991) 517523.
[2] B. Zhou, A. Ilic, On distance spectral radius and distance energy of graphs, MATCH
Commun. Math. Comput. Chem., 64 (2010) 261280
[3] D. M. Cvetkovic, M. Doob, H. Sachs, Spectra of Graphs Theory and Applications, Academic
Press, New York, 1980.
[4] D. Cvetkovic, I. Gutman (Eds.), Applications of Graph Spectra, Math. Inst., Belgrade,
2009.
[5] I. Gutman, The energy of a graph: old and new results, in: A. Betten, A. Kohnert, R.
Laue, A. Wassermann (Eds.), Algebraic Combinatorics and Applications, Springer-Verlag,
100 Kishori P. Narayankar and Lokesh S. B.

Berlin, 2001, pp. 196-211.

[6] I. Gutman, The energy of a graph, Ber. Math. Statist. Sekt. Forsch-ungszentram Graz.,
103 (1978) 122.
[7] I. Gutman, Distance in Thorny graph, Publ. Inst. Math (Beograd), 63(77) (1998) 31-36.
[8] I. Gutman, B. Furtula, H. Hua, Bipartite unicyclic graphs with maximal, second maximal
and third maximal energy, MATCH Commun. Math. Comput. Chem., 58 (2007) 8592 .
[9] I. Gutman, O. E. Polansky, Mathematical Concepts in Organic Chemistry, Springer-Verlag,
Berlin, Chapter 8 (1986).
[10] I. Gutman, M. Medeleanu, On the structure dependence of the largest eigenvalue of the
distance matrix of an Alkane, Indian J. Chem., A37 (1998) 569573.
[11] I. Gutman, S. Zare Firoozabadi, J. A. de la Pea, J. Rada, On the Energy of Regular
Graphs, MATCH Commun. Math. Comput. Chem., 57 (2007) 435442.
[12] H. Hua, On minimal energy of unicyclic graphs with prescribed girth and pendant vertices,
MATCH Commun. Math. Comput. Chem., 57 (2007) 351361.
[13] H. Hua, Bipartite unicyclic graphs with large energy, MATCH Commun. Math. Comput.
Chem., 58, (2007) 5783.
[14] H. Hua, M. Wang, Unicyclic graphs with given number of vertices and minimal energy,
Lin. Algebra Appl., 426 (2007) 478489 .
[15] G. Indulal, I. Gutman, On the distance spectra of some graphs, Math. Commun., 13 (2008)
123131.
[16] G. Indulal, I. Gutman, A. Vijaykumar, On the Distance Energy of a Graph, MATCH
Commun. Math. Comput. Chem., 60 (2008) 461-472.
[17] Kishori P. Narayankar, Lokesh S. B., Peripheral Wiener Index of a Graph, (submitted).
[18] J. Koolen, V. Moulton, Maximal energy graphs, Adv. Appl. Math., 26 (2001) 47-52.
[19] J. Koolen, V. Moulton, Maximal energy bipartite graphs, Graph. Combin., 19 (2003)
131-135.
[20] X. Li, Y. Shi, and I. Gutman, Graph Energy, Springer, New York, NY, USA, 2012.
[21] V. Nikiforov, The energy of graphs and matrices, J. Math. Anal. Appl., 326 (2007) 1472
1475 .
[22] V. Nikiforov, Graphs and matrices with maximal energy, J. Math. Anal. Appl., 327 (2007)
735738 .
[23] S. Ramane, I. Gutman, D. S. Revankar, Distance Equienergetic Graphs, MATCH Commun.
Math. Comput. Chem., 60 (2008) 473484.
[24] H. S. Ramane, D. S. Revankar, I. Gutman, S. B. Rao, B. D. Acharya, H. B. Walikar,
Estimating the distance energy of graphs, Graph Theory Notes N. Y., 55 (2008) 27-32.
[25] H.S. Ramane, H.B. Walikar, S.B. Rao, B.D. Acharya, P.R. Hampiholi, S.R. Jog, I. Gutman,
Spectra and energies of iterated line graphs of regular graphs, Appl. Math. Lett., 18, 679
682 (2005).
[26] H.S. Ramane and H.B. Walikar; Construction of equienergetic graphs, MATCH Commun.
Math. Comput. Chem., 57 (2007) 203210 .
[27] I. Shparlinski, On the energy of some circulant graphs, Lin. Algebra Appl., 414 (2006)
378382 .
 Peripheral Distance Energy of Graphs 101

[28] S. R. Jog, R. Kotambari, Minimum Covering Energy of Some Thorny Graphs, Asian
Journal of Mathematics and Applications, Volume 2014, Article ID ama0171, 7 pages.
[29] W. Yan, L. Ye, On the maximal energy of trees with a given diameter, Appl. Math. Lett.,
18 (2005), 10461052.
[30] L. Ye, X. Yuan, On the minimal energies of trees with a given number of pendent vertices,
MATCH Commun. Math. Comput. Chem., 57 (2007) 193201.
[31] A. Yu, X. Lv, Minimal energy of trees with k pendent vertices, Lin. Algebra Appl., 418
(2006) 625633.
[32] B. Zhou, N. Trinajsti`c, On the largest eigenvalue of the distance matrix of a connected
graph, Chem. Phys. Lett. 447 (2007) 384387.
[33] B. Zhou, I. Gutman, J.A. de la Pea, J. Rada, and L. Mendoza, On spectral moments and
energy of graphs, MATCH Commun. Math. Comput. Chem., 57 (2007) 183191.
Math.Combin.Book Ser. Vol.1(2017), 102-110

Some Properties of a h-Randers Finsler Space

V. K. Chaubey1 , Arunima Mishra2 and A. K. Pandey3

1. Department of Applied Sciences, Buddha Institute of Technology, Gida, Gorakhpur (U.P.)-273209, India

2. Rashtriya Inter College, Baulliya Coloney, Gorakhpur (U.P.)-273001, India

3. Department of Mathematics, Sant Longowal Institute of Engineering and Technology, Sangrur, Punjab, India

E-mail: [email protected], [email protected], [email protected]

Abstract: The purpose of the present paper is to obtain the relation between imbedding
class numbers of tangent Riemannian spaces to (M n , L) and (M n , L ) where L (x, y) is
obtained from the transformation of L(x, y) is given by

L (x, y) L(x, y) + bi (x, y)y i

Key Words: Riemannian metric, h-vector, imbedding class.

AMS(2010): 53B40, 53C60.

1. Introduction

In 1971 Matsumoto [5] introduced the transformation of Finsler metric

y) L(x, y) + bi y i
L(x, (1.1)

and obtain the relation between the imbedding class numbers of a tangent Riemannian spaces
which is obtained by the transformation of the Finsler
to (M n , L) and a Finsler space (M n , L)
metric L by the relation given by in the equation (1.1). Since a concurrent vector field is a func-
tion of (x) i.e., position only, assuming bi (x) as a concurrent vector field, Matsumoto [6] studied
the R3-likeness of Finsler spaces (M n , L) and (M n , L). Singh and Prasad [14,11] generalized
the concept of concurrent vector field and introduced the semi-parallel and concircular vector
fields which are functions of (x) only. Assuming bi (x) as a concicular vector field, Prasad, Singh
and Singh [11] studied the R3-likeness of (M n , L) and (M n , L).
If L(x, y) is a metric function of Riemannian space then L(x, y) reduces to the metric
function of Randers space. Such a Finsler metric was first introduced by G. Randers [13]from
the standpoint of general theory of relativity and applied to the theory of the electron microscope
by R. S. Ingarden [3] who first named it us Randers space. The geometrical properties of this
space have been studied by various workers [2, 7, 9, 12, 15].. In 1970 Numata [10] has studied
which is obtained from Minkowski space (M n , L) by transformation
the properties of (M n , L)
1 Received August 18, 2016, Accepted February 21, 2017.
 Some Properties of a h-Randers Finsler Space 103

(1.1). In all those works the function bi (x) are assumed to be functions of (x) only.
In 1980, Izumi [4] while studying the conformal transformation of Finsler spaces, introduced
the h-vector bi which is v-covariantly constant with respect to Cartans connection C and
satisfies the relation
h
LCij bh = hij

Thus the h-vector bi is not only a function of (x) but it is also a function of directional arguments
satisfying Lj bi = hij . The purpose of the present paper is to obtain the relation between
imbedding class numbers of tangent Riemannian spaces to (M n , L) and (M n , L ) where L (x,
y) is obtained from the transformation of L(x, y) is given by

L (x, y) L(x, y) + (x, y), (1.2)

where (x, y) = bi (x, y)y i , i.e. bi (x, y) is the function of position and direction both.

2. An h-Vector in (M n , L)

Let bi be a vector field in the Finsler space (M n , L). If bi (x, y) satisfies the conditions

bi |j = 0, (2.1)

h
LCij bh = hij , (2.2)

then the vector field bi is called an h-vector [4]. Here |i denotes the v-covariant derivative with
respect to y i in the case of Cartans connection C, Cij h
is the cartans C-tensor, hij is the
angular metric tensor and is given by

LC i bi
= , (2.3)
(n 1)

where C i is the torsion tensor given by Cjk

i jk
g .

Lemma 2.1([4]) If bi is an h-vector then the function and are independent of y.

Since The v-covariant derivation of b2 = g ij bi bj and the fact that g ij is v-covariantly

constant yield
bk b = g ij bi bj |k .

In the view of (2.1) we have

k b = 0.

Thus we have

Lemma 2.2 The magnitude b of an h-vector is independent of y.

r
From (2.1), Ricci identity [8] and the fact that Sihjk = ghr Sijk is skew-symmetric in h and
104 V. K. Chaubey, Arunima Mishra and A. K. Pandey

i we have
h
bi |j |k bi |k |j = Sijk bh = 0.

Thus we have

Lemma 2.3 For an h-vector bi we have Shijk bh = 0, where Shijk are components of v-curvature
tensor of Cartans connection C.

The concept of concurrent vector field in (M n , L) has been introduced by Tachibana [16]
and its properties have been studied by Matsumoto [6]. A vector field bi in (M n , L) is said to
be concurrent if it satisfies the condition (2.1) and

bi|j = gij , (2.4)

where |j denotes h-covariant differentiation with respect to xi in the sense of Cartans connection
C.
Applying Ricci Identity [8]

h h h
bi|j |k bi |k|j = bh Pijk bi|h Cjk bi |h Pjk

and using (2.1) and (2.4) we have

h
Pijk bh + Cijk = 0.

h
Since Pimjk = gmh Pijk is skew-symmetric in i and m, contraction of above equation with
i ij i
b = g bj gives Cijk b = 0. Hence we have the following

Lemma 2.4 An h-vector bi with 6= 0 is not a concurrent vector field.

3. Properties of the h-Randers Finsler Space

Let bi be an h-vector in the Finsler space (M n , L) and (M n , L ) be another Finsler space whose
fundamental function L (x, y) is given by (1.2).
Since bi is an h-vector, from (2.1) and (2.2), we get

j bi = L1 hij , (3.1)

which after using the indicatory property of hij yields j = bj .

Definition 3.1 Let M n be an n-dimensional differentiable manifold and F n be a Finsler space

equipped with a fundamental function L(x, y), (y i = x i ) of M n . A change in the fundamental
function L by the equation (1.2) on the same manifold M n is called h-Randers change. A space
equipped with fundamental metric L is called h-Randers changed Finsler space F n .
 Some Properties of a h-Randers Finsler Space 105

Now differentiating (1.2) with respect to y i we have

li = li + bi , (3.2)

where li = i L is the normalized supporting element in (M n , L) and li = i L is the normalized

element of support in (M n , L ). The quantities of (M n , L ) will be denoted by starred letter.
Now differentiating (3.2) with respect to y j then the angular metric tensor hij = j li is given
by
hij = hij , (3.3)

where = LL1 (1 + ). Hence we have

gij = gij + (1 )li lj + (li bj + lj bi ) + bi bj . (3.4)

From (3.4) the relation between the contravariant components of the fundamental tensors
can be derived as follows

g ij = 1 g ij (1 + 2 ) 3 (1 b2 )li lj (1 + ) 2 (li bj + lj bi ), (3.5)

where b is the magnitude of the vector bi .

From the lemma (2.1) and (3.2) we have

(1 + )
i = mi , (3.6)
L

mi = b i li . (3.7)
L
Now differentiating (3.3) with respect to y k (3.2), (3.6), (3.3) and the fact

k hij = 2Cijk L1 (hik lj + hjk li ),

we have
hij mk + hjk mi + hki mj
Cijk = Cijk + (1 + ) . (3.8)
2L
From the definition of mi , it is evident that

2
(a) mi li , (b) mi bi = b2 L2 = mi mi ,
(3.9)
(c) hij mi = hij bi = mj , (d) Cij
h
mh = L 1
hij .

From (2.1), (3.5), (3.8) and (3.9) we have

r r
(hij mr + hrj mi + hri mj ) 1
Cij = Cij + [{
2L L
L 2 L
+ (b2 2 )}hij + mi mj ]lr . (3.10)
2L L L
106 V. K. Chaubey, Arunima Mishra and A. K. Pandey

Proposition 3.1 Let F n = (M n , L ) be an n-dimensional Finsler space obtained from the

h-Randers change of the Finsler space F n = (M n , L), then the normalized supporting element
li , angular metric tensor hij , fundamental metric tensor gij

and (h)hv-torsion tensor Cijk of
n
F are given by (3.2), (3.3), (3.4) and (3.8) respectively.

Proposition 3.2 Let F n = (M n , L ) be an n-dimensional Finsler space obtained from the

h-Randers change of the Finsler space F n = (M n , L), then the reciprocal of the fundamental

metric tensor gij is given by (3.5).

The curvature tensor Shijk of (M n , L ) is given by

m m
Shijk = Chkm Cij Chjm Cik . (3.11)

From the equation (3.8) and (3.10), we have

m m (1 + )
Chkm Cij = Chkm Cij + hij hhk + {Cijk mh + Chjk mi
2L
(1 + )
+Chik mj + Chij mk } + {2hij mk mh
4LL
+2hhk mi mj + hik mj mh + hih mj mk + hjk mi mh
+hjh mi mk }, (3.12)

(1+) 1+ 2 2
where = 4L2 + 4LL (b L2 ). Thus from (3.11) we have

Shijk = Shijk + hij dhk + hhk dij hik djh hhj dik , (3.13)

1+
where dij = 2 hij + 4LL mi mj .

If we define the tensor Aij and Bij as

hij + dij hij dij

Aij = , Bij = , (3.14)
2 2

then Shijk is written as

Shijk = Shijk (Ahj Aik Ahk Aij ) + (Bhj Bik Bhk Bij ). (3.15)

Thus we have

Proposition 3.3 Let F n = (M n , L ) be an n-dimensional Finsler space obtained from the

h-Randers change of the Finsler space F n = (M n , L), then the curvature tensor Shijk

is given
by (3.15).

If |j denotes v-covariant differentiation with respect to y j in (M n , L ) then we have

(hij lk hik lj )
hij |k hik |j = , (3.16)
L
 Some Properties of a h-Randers Finsler Space 107

(mi lj mj li )
mi |j mj |i = , (3.17)
L
(dij lk dik lj )
dij |k dik |j = . (3.18)
L
Hence from (3.14), (3.16) and (3.18), we get

(Bij lk Bik lj )
Aij |k Aik |j = , (3.19)
L

(Aij lk Aik lj )
Bij |k Bik |j = . (3.20)
L

4. Imbedding Class Numbers of Tangent Riemannian Space to (M n , L) and (M n , L )

The tangent vector space Mxn to M n at every point x is regarded as n-dimensional Riemannian
space (Mxn , gx ) with Riemannian metric gx = gij (x, y)dy i dy j . Thus the component Cjk
i
of
Cartans C-tensor are the Christoffel symbols associated with gx , i.e.

1 ih
i
Cjk = g (k gjh + j ghk + h gjk ).
2
i
Hence Cjk defines the Riemannian connection on Mxn . It is observed from the definition if Shijk
that the curvature tensor of the Riemannian space (Mxn , gx ) at a point x. The space (Mxn , gx )
equipped with such a Riemannian connection will be called the tangent Riemannian space.

In the theory of Riemannian space, we know that any n-dimensional Riemannian space
V , can be imbedded isometrically in a Euclidean space of dimension n(n1)
n
2 . If n + r is the
lowest dimension of the Euclidean space in which V n is imbedded isometrically then the integer
r is called imbedding class number of V n . The fundamental theorem of isometric imbedding
[1] states that the tangent Riemannian n-space (Mxn , gx ) is locally imbedded isometrically in
an Euclidean n + r space if and only if there exist r numbers, and = 1, r symmetric tensor
H(P )ij and r(r1)
2 covariant vector fields H(P Q)i = H(QP )i , Q = 1, 2, 3, , r satisfying the
Gauss equations,

Shijk = Sigma(P ) {H(P )hj H(P )ik H(P )hk H(P )ij },

where summation is given over P .

The Codazzi equations

H(P )ij |k H(P )ik |j = (Q) {H(Q)ij H(QP )k H(Q)ik H(QP )j },

where summation is given over Q and Ricci-Kuhne equations

H(P Q)i |j H(P Q)j |i + (R) {H(RP )i H(RQ)j

H(RP )j H(P Q)i } + g hk {H(P )hi H(Q)kj H(P )hj H(Q)ki } = 0.
108 V. K. Chaubey, Arunima Mishra and A. K. Pandey

For a special case when (Mxn , gx ) is of imbedding class 1, the above equations reduce to

Shijk = (Hhj Hik Hhk Hij ), (4.1)

Hij |k Hik |j = 0. (4.2)

Since Shijk y k = 0, from (3.21), we have

Hhj Hi0 Hh0 Hij = 0

contracting above equation by y i , we have

Hhj H00 Hh0 H0j = 0,

1
which implies that H0j = 0 or Hij = H00 Hh0 H0j . In the latter case we get Shijk = 0. In the
theory of spaces of imbedding class 1, [17] introduced the concept of type number t, which is
the rank of matrix k Hij k provided to the rank is more than 1. If the rank is 0 or 1, then
S vanishes. Therefore if (Mxn , gx ) is of imbedding class 1, the second fundamental tensor Hij
satisfies Hij y j = 0 and thus the type number t is less than n.
Again by virtue of Lemma 2.3 and equation (4.1), we get

Hhj Hik Hhk Hij bh = 0.

From this equation we have

Hhj bh bj Hik Hhk bh Hij bj = 0.

This gives
Hhk bh Hij bj
Hhk bh = 0, or Hik = .
Hhj bh bj

In the latter case Shijk = 0. Thus for an imbedding class 1, Hhk bk = 0. Now we shall put

H(1)ij = Hij , 1 = , (4.3)

H(2)ij = Aij , 2 = 1, 4.4

H(3)ij = Bij , 3 = 1, 4.5

then from (3.15) and (4.1), we get

Shijk = P {H(P

)hj H(P )ik H(P )hh H(P )ij },

where summation is varies from P = 1, 2, 3. Thus the above equation is noting but Gauss
equation of (Mxn , gx ).
Now we put

H(21)i = H(12)i = 0, (4.6)
 Some Properties of a h-Randers Finsler Space 109

H(31)i = H(13)i = 0, (4.7)

1
H(32)i = H(23)i = li (4.8)
L
and using (4.2), (4.3), (3.3), Lemma 2.1 and the fat that Hi0 = 0, we get

H(1)ij |k H(1)ik |j = 0. (4.9)

Again in view of (4.4), (4.5), (4.6), (4.7) and (4.8), equations (3.19) and (3.20) reduce to

H(2)ij |k H(2)ik |j = Q {H(Q)ij

H(Q2)k
H(Q)ik
H(Q2)j }, 4.10

H(3)ij |k H(3)ik |j = Q {H(Q)ij

H(Q3)k
H(Q)ik
H(Q3)j }, 4.11

where summation is varies from Q = 1, 2, 3.

The equations (4.9), (4.10) and (4.11) are the Codazzi equations of (Mxn , gx ). Now we have
to verify Ricci-Kuhne equations, we have from (3.10),

2
li |j = L1 hij+L1 [{ + (2L )1 (v 2 )}hij + L1 mi mj ]
L2
from which we get li |j lj |i = 0. Hence from (4.10), we get

H(32)i |j H(23)j |i = 0,

which are the Ricci-Kuhne equations of (Mxn , gx ) as

M(12) M(21) = 0, and M(13) M(31) = 0.

Thus from above we have

Theorem 4.1 Let F n = (M n , L ) be an n-dimensional Finsler space obtained from the h-

Randers change of the Finsler space F n = (M n , L), then if the tangent Riemannian n-space
(Mxn , gx ) to (M n , L) is of imbedding class 1, then the tangent Riemannian n-space (Mxn , gx ) to
(M n , L ) is at most of imbedding class 3.

References

[1] Eisenhart L. P., Riemannian Geometry, Princeton (1925).

[2] Hashiguchi M. and Ichijyo Y., On some special (, ) metrics, Rep. Fac. Sci., Kagoshima
Univ., 8 (1975), 39-46.
[3] Ingarden R. S., Differential geometry and physics, Tensor, N.S., 20 (1970), 201-209.
[4] Izumi H., Conformal transformations of Finsler spaces II. An h-conformally flat Finsler
space, Tensor, N.S., 33 (1980), 337-359.
[5] Matsumoto M., On transformations of locally Minkowskian space, Tensor, N.S., 22 (1971),
103-111.
110 V. K. Chaubey, Arunima Mishra and A. K. Pandey

[5] Matsumoto M., Finsler space admitting concurrent vector field, Tensor, N.S., 28 (1974),
239-249.
[6] Matsumoto M., On Finsler spaces with Randers metric and special forms of important
tensors, J. Math. Kyoto Univ., 14 (1975), 477-498.
[7] Matsumoto M., Foundations of Finsler Geometry and Special Finsler Spaces, Kaiseisha
Press, Saikawa, Japan (1986).
[8] Numata S., On the curvature tensor Shijk and the tensor Thijk of generalized Randers
spaces, Tensor, N.S., 20 (1975), 35-39.
[9] Numata S., On the torsion tensor Rhjk and Phjk of Finsler spaces with metric ds =
(gij dxi dxj ) + bi (x)dxi , Tensor, N. S., 32 (1978), 27-31.
p

[10] Prasad B. N., Singh V. P. and Singh Y. P., On concircular vector fields in Finsler space,
Indian J. Pure Appl. Math., 17 (1986), 998-1007.
[11] Pandey T. N. and Chaubey V. K., mth-root Randers change of a Finsler Metric, Interna-
tional J. Math. Combin., 1, (2013), 38-45.
[12] Randers G., On an asymmetrical metric in the four space of general relativity, Phys. Rev.,
(2) 59 (1941), 195-199.
[13] Singh U. P. and Prasad B. N., Modification of a Finsler space by a normalized semi-parallel
vector field, Periodica Mathematica Hungarica, 14 (1) (1983), 31-41.
[14] Shibata C., Shimada H., Azumi, M. and Yasuda, H., On Finsler spaces with Randers
metric, Tensor, N. S., 31 (1977), 219-226.
[15] Tachibana S., On Finsler spaces which admit a concurrent vector field, Tensor, N. S. 1
(1950), 1-5.
[16] Thomas T. Y., Riemannian spaces of class one and their characterization, Acta Math.,67
(1936), 169-211.
Math.Combin.Book Ser. Vol.1(2017), 111-126

Pure Edge-Neighbor-Integrity of Graphs

Sultan Senan Mahde and Veena Mathad

(Department of Studies in Mathematics, University of Mysore, Manasagangotri, Mysuru, India)

E-mail: [email protected], veena [email protected]

Abstract: In a communication network, several vulnerability measures are used to de-

termine the resistance of the network to disruption of operation after the failure of certain
stations or communication links. This study introduces a new vulnerability parameter, pure
edge-neighbor-integrity of graphs. The pure edge-neighbor-integrity of a graph G is defined
to be P EN I(G) = minE(G) {|| + e (G/)}, where is any edge subversion strategy of
G and e (G/) is the number of edges in the largest component of G/. A set E(G), is
said to be a P EN I-set of G if P EN I(G) = || + e (G/). In this paper, several properties
and bounds on the P EN I are presented over here and the relation between P EN I with
other parameters is investigated. The P EN I of some classes of graphs is also computed.

Key Words: Vulnerability, integrity, neighbor-integrity, edge-neighbor-integrity.

AMS(2010): 05C40, 05C99, 05C76.

1. Introduction

Networks appear in many different applications and settings. The most common networks are
telecommunication networks, computer networks, the internet, road and rail networks and other
logistic networks. In all applications, vulnerability and reliability are crucial and important
features. Network designers often build a network configuration around specific processing,
performance and cost requirements. But there is little consideration given to the stability of
the networks communication structure when under the pressure of link or node loses. This lack
of consideration makes the networks have low survivability. Therefore, network design process
must identify the critical points of failure and be able to modify the design to eliminate them
[18].
A network can be modeled by a graph whose vertices represent the stations and whose edges
represent the communication lines. Vulnerability measures the resistivity of the network to the
disruption of its operation due to the failure of certain stations or communication links. Losing
links or nodes eventually lead to a loss of the effectiveness of the network. Communication
networks must be constructed so as to be as stable as possible, not only with respect to the initial
disruption, but also with respect to the possible reconstruction of the network. Many graph
1 Received April 7, 2016, Accepted February 25, 2017.
112 Sultan Senan Mahde and Veena Mathad

theoretical parameters have been used in the past to describe the stability of communication
networks, including connectivity, integrity, toughness and binding number. However, these
parameters do not take into account the effect that the removal of a vertex has on the neighbors
of that vertex. If a station is destroyed, the adjacent stations are betrayed and become useless
to the network as a whole. The neighbor integrity is a measure of the vulnerability of graphs to
the disruption caused by the consecutive removal of a vertex and all of its adjacent vertices [8,
9, 10, 15] a probabilistic basis. However, sometimes it is important to take subjective reliability
estimates into consideration. Among the relevant issue of importance, we are particularly
interested in one of the vulnerabilities. That is, in an unfriendly external environment, how
vulnerable is such a distributed system to certain external destruction and how much computing
power can be sustained in the face of destruction.
The concept of network vulnerability is motivated by the design and analysis of networks
under a hostile environment. Several graph theoretic models under various assumptions have
been proposed for the study and assessment of network vulnerability. Graph integrity, intro-
duced by Barefoot et al. [4, 5], is one of these models that has received wide attention [2,
11].
In 1994, Margaret B. Cozzens and Wu [7] introduced a new graph parameter called the edge-
neighbor-integrity. They consider the edge analogue of (vertex )neighbor-integrity a measure
of the vulnerability of graphs to disruption caused by the removal of edges, their incident
vertices, and all of their incident edges. The integrity of a graph G = (V, E), which was
introduced as a useful measure of the vulnerability of the graph, is defined as follows: I(G) =
min{|S| + m(G S) : S V (G)}, where m(G S) denotes the order of the largest component.
Barefoot, Entringer and Swart defined the edge-integrity of a graph G with edge set E(G) by
I (G) = min{|S| + m(G S) : S E(G)}. The weak integrity was introduced by Kirlangic
[14] and is defined as Iw (G) = min{|S| + me (G S) : S V (G)}, where me (G S) denotes
the number of edges in a largest component of G S. Let u be a vertex in G. N (u) = {v
V (G)|u 6= v, v and u are adjacent} is the open neighbourhood of u, and N [u] = {u} N (u)
denotes the closed neighborhood of u. A vertex u in G is said to be subverted if the closed
neighborhood N (u) is deleted from G. A set of vertices S = {u1 , u2 , , un } is called a vertex
subversion strategy of G if each of the vertices in S has been subverted from G. Let G/S be the
survival-subgraph when S has been a vertex subversion strategy of G. The closed neighborhood
of a vertex subset S, N [S], is uS N [u]. Hence G/S = G N [S] = G (uS N [u]). The
vertex-neighbor-integrity of a graph G, V N I(G), is defined to be V N I(G) = minSV (G) {|S| +
(G/S)}, where S is any vertex subversion strategy of G, and (G/S) is the maximum order
of the components of G/S. The edge e = (v, w) in G is said to be subverted if the edge e,
all of its incident edges, and the two ends of e, namely v and w, are removed from G. (For
simplicity, an edge e = (v, w) is subverted if the two ends of the edge e, namely v and w, are
deleted from G.) A set of edges = {e1 , e2 , , en } is called an edge subversion strategy of G
if each of the edges in has been subverted from G. Let G/ be the survival-subgraph when
has been an edge subversion strategy of G. The edge-neighbor-integrity of a graph G, is
defined to be EN I(G) = minE(G) {|| + (G/)}, where is any edge subversion strategy
of G, and (G/) is the maximum order of the components of G/. We now introduce
 Pure Edge-Neighbor-Integrity of Graphs 113

a new measure of stability of a graph G in this sense and it is called pure edge-neighbor-
integrity. Formally, the pure edge-neighbor-integrity P EN I(G) of a graph G is defined as
P EN I(G) = minE(G) {|| + e (G/)}, where is any edge subversion strategy of G and
e (G/) is the number of edges of a largest component of G/. Any set with property that
P EN I(G) = || + e (G/) is called a P EN I set of G. x is the smallest integer greater
than or equal to x. x is the greatest integer less than or equal to x.
By a graph G = (V, E), we mean a finite undirected graph without loops or multiple
edges, with vertex set V (G) = {v1 , v2 , , vp }. The distance between the vertices vi and vj
is the length of the shortest path joining vi and vj . The shortest vi vj path is often called a
geodesic. The diameter of a connected graph G is the length of any longest geodesic, denoted
by diam(G). The order and size of G are denoted by p and q, respectively. We use Bondy
and Murty [6, 12] for terminology and notations not defined here. In general, the degree of a
vertex v in a graph G is the number of edges of G incident with v and it is denoted by degv.
The maximum (minimum) degree among the vertices of G is denoted by (G), ((G)). We
denote the minimum number of edges in edge cover of G ( i.e., edge cover number ) by 1 (G)
and the minimum number of edges in independent set of edges of G (i.e., edge independence
number) by 1 (G). A vertex of degree one is called a pendant vertex. The symbols (G), (G),
(G), and (G) denote the vertex cover number, the connectivity, the edge-connectivity, and
the independence number of G, respectively.
A subset X of E is called an edge dominating set of G if every edge not in X is adjacent
to some edge in X. The edge domination number (G) of G is the minimum cardinality taken
over all edge dominating sets of G [16].
The line graph L(G) of G has the edges of G as its vertices which are adjacent in L(G)
if and only if the corresponding edges are adjacent in G [12]. In the present work, the basic
properties of pure edge-neighbor-integrity and of P EN Isets are explored, bounds and rela-
tionship between pure edge-neighbor-integrity and other graphical parameters are considered.
Finally, the pure edge-neighbor-integrity of binary operations of some graphs are determined.
We need the following to prove main results.

Lemma 1.1([13]) If D E(G), then L(G D) = L(G) D.

Theorem 1.1([14]) If a graph G of order n is isomorphic to a cycle graph or a tree, then

Iw (G) = I(G) 1.

Theorem 1.2([12]) For any graph G, (G) (G) (G).

Lemma 1.2 For any graph G, 1 (G) (G).

p
Theorem 1.3([1]) For any connected graph G of even order p, = 2 if and only if G is
isomorphic to Kp or K 2p , p2 .

Theorem 1.4([2]) The integrity of

(a) the complete graph Kp is p;

(b) the complete bipartite graph Km,n is 1 + min{m, n}.
114 Sultan Senan Mahde and Veena Mathad

2. Main results

Proposition 2.1 (a) For any complete graph Kp , P EN I(Kp ) = p2 ;

(b) For any path Pp with p 3, P EN I(Pp ) = 2 p + 2 4;
(c) For any cycle Cp ,

1, if p = 3 ;


P EN I(Cp ) = 2, if p = 4 ;



2 p 3, if p 5.

(d) For the star K1,p1 , P EN I(K1,p1 ) = 1;

(e) For the double star Sn,m , P EN I(Sn,m ) = 1;
(f ) For the complete bipartite graph Kn,m , P EN I(Kn,m ) = min{n, m};

(g) For the wheel graph W1,p1 , p 5, P EN I(W1,p1 ) = 2 p 3.

Remark 2.1 (1) If H is a subgraph of G, then P EN I(H) P EN I(G);

(2) Pure edge-neighbor integrity of a connected graph for p 2, takes its minimum value
at K1,p1 and its maximum value at Kp complete graph;
(3) 0 P EN I(G) q.

Lemma 2.1 If G is a non-trivial graph, then for all v V (G), P EN I(G v) P EN I(G) 1,
the bound is sharp for G = K4 .

Proposition 2.2 (a) If G has enough components close in size to the largest one, then
P EN I(G) = e (G). In particular, if G = pH with p e (H), then P EN I(G) = e (H);
(b) Suppose that G is disconnected and m(G) = k, if G has at least k 1 components of
order k, then empty set is an P EN I(G)-set of G.

Lemma 2.2 If is P EN I-set of G, then e (G/) = P EN I(G/) and is P EN I-set of

G/.

Proof Let is P EN I-set of G and be P EN I-set of G/. Thus

|| + e (G/) = P EN I(G)
e (G/( )) + | |
= || + e [(G/)/ ] + | |
= || + P EN I(G/).

So, e (G/) P EN I(G/), but e (G/) P EN I(G/). This completes the proof. 2

Lemma 2.3 If D E(G),P EN I(L(G D)) = P EN I(L(G) D)).

Proof The proof follows by Lemma 1.1. 2

Theorem 2.1 If G is a simple graph such that G
= L(G), then P EN I(G) = P EN I(L(G)) =
 Pure Edge-Neighbor-Integrity of Graphs 115

P EN I(G) if and only if G = C5 or G is the graph shown in the Figure 1.

Figure 1 G

Proposition 2.3 If a connected graph G is isomorphic to its line graph, then P EN I(G) =
P EN I(L(G)). But the converse is not true, for example the graph G is given in the following
Figure 2.

G L(G)
Figure 2 G and L(G)

Notice that P EN I(G) = 2 = P EN I(L(G)), but G and L(G) are not isomorphic.

Lemma 2.4 Let G be a connected graph of order at least 3. If P EN I(G) = 1, then the diameter
of G is 3.

Proof The diameter of G is 4 is Supposed, then G contains a path P5 . Hence for any
edge e in G, e (G/e) 1, and for any two edges e1 and e2 in G, e (G/{e1 , e2 }) 0. Thus
EEN I(G) 2, a contradiction. Hence, the diameter of G is 3. 2
Lemma 2.5 For any a graph G, P EN I(G) = V N I(L(G)).

Proof Since every edge dominating set in G is a dominating set in the line graph of G, the
set of edges S that satisfies P EN I(G) equal to the set of vertices S that satisfies V N I(L(G)),
this completes the proof. 2
Lemma 2.6 For any (p, q) graph G, q+1 P EN I(G) q 1 , where denotes the
maximum degree of an edge in G.

Observation 2.1 For any connected graph G, P EN I(G) = q 1 if and only if G

= Pp , 3

p 6, G = p8 , G = C4 or G in the Figure 3.

Figure 3 G
116 Sultan Senan Mahde and Veena Mathad

Corollary 2.1 For any connected (p, q) graph, P EN I(G) = pq if and only if G is isomorphic
to K1,p1 or Sn,m .

Observation 2.2 Let G be a graph, and let be P EN I-set of G such that || = 1, then the
following hold

(a) P EN I(G) = 1;
P
(b) |E | = e deg(e);

(c) (G) = q 1.
p
Corollary 2.2 For any connected graph G of even order p, P EN I(G) = 2 if and only if G is
isomorphic to Kp or K 2p , p2 .

Theorem 2.2 For any integer n 1, there does not exist any graph G satisfy P EN I(G) =
I(G) = (G) = n.

Proof Let G be a graph of order p. By Theorem 1.3 and Corollary 2.2, P EN I(G) =
p
2 = (G) if p is even and G = Kp or G
= K p2 , p2 , but from Theorem 1.4, I(Kp ) = p, and
p
I(K p2 , p2 ) = 2 + 1. Hence the result. 2
Theorem 2.3 For any integer k 1, there exists a graph G of size q k with P EN I(G) =
(G) = k, where (G) is domination number.

Proof The result is true for k = 1, 2, since G1 = K2 , G2 = K3 have the desired property.
For k 3, consider the graph Gk which is obtained from k disjoint copies of the complete graph
K3 and joining the vertex vi in the ith copy with the vertex vi+1 in the (i + 1)th copy, and
joining the vertex ui in the ith copy with the vertex wi in the (i + 1)th copy. The graph G3
shown in Figure 4.
v1 v2 v3

v u1 w1 u2 w2 w

Figure 4 G3

Consider D = {v1 , v2 , v3 , , vk } be a dominating set for Gk , and |D| = k. Lets claim that
set D is a minimum dominating set. Since each vi , 2 i k 1, is adjacent to wi1 and ui . If
vi is removed from set D, then wi1 and ui will not be dominated by any vertex. Hence D is a
minimum domination set. Therefore, (Gk ) = k. Consider = {(v1 , v), (v2 , w1 ), (v3 , w2 ), ,
(vi , wi1 ), 1 i k}. Then || = k, and e (Gk /) = 0. Therefore, P EN I(Gk ) || +
e (Gk /) = k. Consider 1 = {(v1 , v), (v2 , w1 ), (v3 , w2 ), ..., (vi1 , wi2 ), 1 i k}. Then
|1 | = k 1, and e (Gk /1 ) = 4, this implies that |1 | + e (Gk /1 ) > || + e (Gk /). If
e (Gk /) = 1, then || k. Thus, P EN I(Gk ) k + 1. Therefore, P EN I(Gk ) = k. 2
Corollary 2.3 For every integer n 1, there exists graph G with P EN I(G) = n.
 Pure Edge-Neighbor-Integrity of Graphs 117

Lemma 2.7 Let G be a graph of order p, P EN I(G) = 0 if and only if G

= K p.

Theorem 2.4 For any graph G of order p, P EN I(G) Iw (G) I (G).

Proof Clearly, Iw (G) I (G). If G is complete, then P EN I(G) = p2 |p 1| = Iw (G).

G is non-complete is supposed and S = {u1 , u2 , , up } be an Iw -set of G. Then S is a vertex
cut-set of G, and ui , where 1 i p, is not an isolated vertex of G. Let = {(ui , vi ) E(G)/
for some vertex vi V, ui S , where i = 1, 2, , p} thus || = |S | = p. Therefore,

G/ = G {u1 , u2 , , up , v1 , v2 , , vp }
= G (S {vi V (G)/(ui , vi ) , ui S }) G S ,

it follows that e (G/) me (G S ), then

P EN I(G) || + e (G/)
|S | + me (G S ) = Iw (G). 2
Observation 2.3 If P EN I(G) = Iw (G), then the induced subgraph of G, < S > must be a
null graph, where S is an Iw -set of G. But the converse is not true, for example in the graph
in Figure 5. S = {u1 , u2 , u3 } is an Iw -set of G is noted. Therefore, Iw (G) = 4 and = {e1 , e2 }
is a P EN I- set of G. Thus P EN I(G) = 2. < S > is a null graph, but Iw (G) 6= P EN I(G).
u1

u2

e2 e1
u3

Figure 5

Lemma 2.8 If diam(L(G)) = 1, then P EN I(G) = 1.

Proof Since diam(L(G)) = 1, then G is either K3 or K1,p1 . Hence the result. 2

Remark 2.2 If G is a graph with (G) = 1, P EN I(L(G)) = p2 .

Theorem 2.5 For any graph G, V N I(G) P EN I(G) + 1.

Proof Let = {(u1 , v1 ), (u2 , v2 ), , (un , vn )} be a P EN I-set of G, let S be a set of

one end vertex of each edge in . Thus |S| || and {u1 , u2 , , un , v1 , v2 , , vn } N [S].
Therefore G/S = G N [S] G {u1 , u2 , , un , v1 , v2 , , vn } = G/, and |S| + (G/S)
||+(G/) ||+e (G/)+1 = P EN I(G)+1. So V N I(G) = minSV (G) {|S|+(G/S)}
|S | + (G/S ) P EN I(G) + 1. 2
118 Sultan Senan Mahde and Veena Mathad

Theorem 2.6 For any graph G, P EN I(G) EN I(G) 1.

Proof Let be a P EN I-set of G. Since EN I(G) || + (G/) and (G/)

e (G/) + 1, for every E(G), hence the result. 2
Theorem 2.7 For any graph G and e E(G), P EN I(G e) P EN I(G) 1.

Proof Let be a P EN I-set of G e, and P EN I(G e) = | | + e ((G e)/ ),

let = {e}. Then || = | | + 1. Then is P EN I-set of G and e (G/ ) =
e (G/e)/ ) Therefore,

P EN I(G) | | + e (G/ )
| | + e [(G/e)/ ] + 1
= P EN I(G e) + 1.

Then P EN I(G e) P EN I(G) 1. 2

Theorem 2.8 For any graph G, P EN I(G) 1 (G).

Proof Let be P EN I-set of G such that P EN I(G) = || + e (G/) and let C be a

minimum edge covering of G. Since each vertex of G is an end vertex of some edge in C, we
have G/C = and e (G/C) = 0.
Thus

P EN I(G) = || + e (G/)
|C| + e (G/C) = |C| = 1 (G). 2
Theorem 2.9 For any graph G, P EN I(G) 1 (G).

Proof Let be P EN I-set of G such that P EN I(G) = || + e (G/) and let M be a

maximum matching in G. It is clear G/M = or a set of isolated vertices, hence e (G/M ) = 0.
Then

P EN I(G) = || + e (G/)
|M | + e (G/M ) = |M | = 1 (G). 2
Theorem 2.10 For any graph G, P EN I(G) (G).

Proof The proof follows from Lemma 1.2 and Theorem 2.9. 2
Theorem 2.11 For any tree T , P EN I(T ) (T ).

Proof Let be a P EN I-set of T such that P EN I(T ) = || + e (T /). Then e (T /)

(T /) (T ) ||, So, P EN I(T ) = || + e (T /) || + (T ) || = (T ). 2
Lemma 2.9 For any tree T , P EN I(T ) (T ).
 Pure Edge-Neighbor-Integrity of Graphs 119

Proof The proof follows from Theorems 2.11 and 1.2. 2

Lemma 2.10 For any tree T , P EN I(T ) (T ).

Proof The proof follows from Lemma 2.9 and Theorem 1.2. 2
Notice that 1 (G), 1 (G) and (G) are upper bounds of P EN I(G), while (G), (G) and
(G) are lower bounds of P EN I(G).
However, the independence number , has no such relationship with P EN I(G). For example,

(1) P EN I(K1,n ) < (K1,n );

(2) P EN I(Kp ) > (Kp );

n = m = (K ), if n = m ;
n,m
(3) P EN I(Kn,m ) =
min{n, m} < (Kn,m ), if n 6= m.

Corollary 2.4 For any graph G, P EN I(G) p2 .

Proof Let M be a maximum matching of G. Then |M | = 1 (G) p2 . Two cases are

discussed.

Case 1. If 1 (G) = p2 , then G/M = (if p is even) or a single vertex (if p is odd), hence
P EN I(G) |M | + e (G/M ) = p2 .

Case 2. If 1 (G) < p2 , then by Theorem 2.9, we have P EN I(G) 1 (G) < p2 . 2
Theorem 2.12 For any graph G, P EN I(G) I(G)
2 1.

Proof Let = {(u1 , v1 ), (u2 , v2 ), , (un , vn )} be a P EN I-set of G. So P EN I(G) =

|| + e (G/) = n + e (G/).
Let S = {u1 , u2 , , un , v1 , v2 , , vn }. Since may not be edge independent in G,
|S | 2n. Then

I(G) = minSV (G) {|S| + m(G S)}

|S | + m(G S ) 2n + e (G/) + 1
2(n + e (G/)) + 1 = 2P EN I(G) + 1.

Therefore, P EN I(G) I(G)

2 1. 2
Corollary 2.5 For any graph G, P EN I(G) Iw2(G) .

3. Pure-Edge Neighbor Integrity of Some Graph Operators

Definition 3.1([12]) The (Cartesian)product G H of graphs G and H has V (G) V (H) as

its vertex set and (u1 , u2 ) is adjacent to (v1 , v2 ) if either u1 = v1 and u2 is adjacent to v2 or
u2 = v2 and u1 is adjacent to v1 .
120 Sultan Senan Mahde and Veena Mathad

Theorem 3.1 For a graph K2 Pp ,

p
2 + 1, p is even;

P EN I(K2 Pp ) =
p1

2 + 1, p is odd.

Proof The number of vertices of graph K2 Pp is 2p and the number of edges is 3p 2.

The graph K2 Pp is shown in Figure 6, we have two cases.

Case 1. p is even. Consider = {e2+2j , 0 j < p2 }, || = p2 , and e ((K2 Pp )/) = 1.

Therefore,
p
P EN I(K2 Pp ) || + e ((K2 Pp )/) = + 1. (1)
2
If is set of any edges such that e ((K2 Pp )/) = 0, then || p 1. So

P EN I(K2 Pp ) p 1. (2)

If is set of any edges such that e ((K2 Pp )/) 2, then || p2 , p > 2. Thus

p
P EN I(K2 Pp ) + 2. (3)
2
p
Therefore, the inequalities (1), (2) and (3) lead to P EN I(K2 Pp ) = 2 + 1.

p1 p1
Case 2. p is odd. Consider = {e2+2j , 0 j < 2 }, || = 2 , and e ((K2 Pp )/) = 1.
Therefore,
p1
P EN I(K2 Pp ) || + e ((K2 Pp )/) = + 1. (4)
2
If is set of any edges such that e ((K2 Pp )/) = 0, then || p 1. So

P EN I(K2 Pp ) p 1. (5)

p1
If is set of any edges such that e ((K2 Pp )/) 2, then || 2 + 1. Thus

p1
P EN I(K2 Pp ) > + 1. (6)
2

Therefore, these inequalities (4), (5) and (6) lead to P EN I(K2 Pp ) = p1

2 + 1. 2

u1 u2 u3 up1 up

e1 e2 e3 ep1 ep

v1 v2 v3 vp1 vp

Figure 6 K2 Pp
 Pure Edge-Neighbor-Integrity of Graphs 121

Theorem 3.2 For a graph K2 Cp ,

p
2 + 1, p is even and p > 2;

P EN I(K2 Cp ) =
p+1

2 + 1, p is odd and p 3.

Proof The number of vertices of graph K2 Cp is 2p and the number of edges is 3p. The
graph K2 Cp is shown in Figure 7, two cases are considered.

Case 1. p is even. Consider = {e1+2j , 0 j < p2 }, || = p2 , and e ((K2 Cp )/) = 1.

Therefore,
p
P EN I(K2 Cp ) || + e ((K2 Cp )/) = + 1. (7)
2
If is set of any edges such that e ((K2 Cp )/) = 0, then || p 1. So

P EN I(K2 Cp ) p 1. (8)

p
If is set of any edges such that e ((K2 Cp )/) 2, then || 2 + 1. Thus

p
P EN I(K2 Cp ) + 3. (9)
2
Therefore, these inequalities (7), (8) and (9) lead to
p
P EN I(K2 Cp ) = + 1.
2
Case 2. (i) p is odd, p = 3. Consider S = {e1 , e2 }, |S| = 2, and e ((K2 Cp )/)) = 1. Thus,

P EN I(K2 Cp ) || + e ((K2 Cp )/) = 3.

p+1 p+1
(ii) p > 3, Consider = {e1+2j , 0 j < 2 }, || = 2 and e ((K2 Pp )/) = 1.
Therefore,
p+1
P EN I(K2 Cp ) || + e ((K2 Cp )/) = + 1. (10)
2
If is set of any edges such that e ((K2 Cp )/) = 0, then || p 1. So

P EN I(K2 Cp ) p 1. (11)

p+1
If is set of any edges such that e ((K2 Cp )/) 2, then || 2 . Thus

p+1
P EN I(K2 Cp ) + 2. (12)
2
Therefore, these inequalities (10), (11) and (12) lead to

p+1
P EN I(K2 Cp ) =
2
+ 1. 2
122 Sultan Senan Mahde and Veena Mathad

e1 e2 e3 ep1 ep

Figure 7 K2 Cp

Theorem 3.3 P EN I(K2 K1,p1 ) = 2.

Proof The number of vertices of graph K2 K1,p1 is 2p. The set = {e} as shown in
Figure 8 is chosen. If we remove the edge e, p 1 components such that e ((K2 K1,p1 )/) =
1, thus || + e ((K2 K1,p1 )/) = 2. Therefore, P EN I(K2 K1,p1 ) = 2. If is set of any
edges such that e ((K2 K1,p1 )/) = 0, then || p 1. So P EN I(K2 K1,p1 ) p 1.
If ((K2 K1,p1 )/) 2, then trivially || + e ((K2 K1,p1 )/) > 2. Thus HI(K2
K1,p1 ) = 2. 2

Figure 8 K2 K1,p1

Definition 3.2([12]) For a simple connected graph G the square of G denoted by G2 , is defined
as the graph with the same vertex set as of G and two vertices are adjacent in G2 if they are at
a distance 1 or 2 in G.

Theorem 3.4 For a graph Pp2 ,

p
3,

if p 0(mod 3),
P EN I(Pp2 ) = p1
3 , if p 1(mod 3),

p2

3 + 1, if p 2(mod 3).

Proof Let V (Pp ) = {v1 , v2 , , vp }. Then, |V (Pp2 )| = p and |E(Pp2 )| = 2p 3. The graph
P52 is shown in Figure 9.
 Pure Edge-Neighbor-Integrity of Graphs 123

v1 e1 v2 e2 v
3
e3 v4 e4 v5

Figure 9 P52

An edge set of Pp2 as below is considered.

(1) If p 0(mod 3), then p = 3k for some integer k 1. Consider

= {e2+3i /0 i k 1} and || = k.

We have, || = p3 , and e (Pp2 /) = 0;

(2) If p 1(mod 3), then p = 3k + 1 for some integer k 1. Consider

= {e2+3i /0 i k 1} and || = k.

We have, || = p1 2
3 , and e (Pp /) = 0;
(3) If p 2(mod 3) then, p = 3k 1 for some integer k 1. Consider

= {e1+3i /0 i k 1} and || = k.

p2
We have, || = 3 + 1 and e (Pp2 /) = 0.
To discuss the minimality of || + e (Pp2 /). Consider any edge set 1 of Pp2 such that,
|1 | ||, then due to the construction of Pp2 (i.e., to convert Pp2 /1 into disconnected graph,
include at least one edge in 1 ) must be included. It generates a large value of e (Pp2 /1 ) such
that,
|| + e (Pp2 /) |1 | + e (Pp2 /1 ) (13)

Let 2 be any edge set of Pp2 such that e (Pp2 /2 ) 1. Then

|| + e (Pp2 /) |2 | + e (Pp2 /2 ). (14)

Therefore, these inequalities (13) and (14) lead to

|| + e (Pp2 /) = min{|X| + e (G/X) : X E(G)} = P EN I(Pp2 ). 2

Definition 3.3([17]) The lollipop graph Lp,d is obtained from a complete graph Kpd and a
path Pd , by joining one of the end vertices of Pd to all the vertices of Kpd .

Theorem 3.5 For a lollipop graph Lp,d ,

pd+1
P EN I(Lp,d ) = + 2 d + 1 4.
2

Proof The number of the vertices of Lp,d is p and the number of edges is d1+ (pd+1)(pd)
2 .
124 Sultan Senan Mahde and Veena Mathad

The graph Lp,d consists of a complete graph of order p d + 1 and a path of order d 1. By
Proposition 2.1, it follows that

pd+1
P EN I(Lp,d ) = P EN I(Pd1 ) + P EN I(Kpd+1 ) =
2
+ 2 d + 1 4. 2
Definition 3.4([17]) A broom graph Bp,d consists of a path Pd , together with (p d) end
vertices all adjacent to the same end vertex of Pd .

Theorem 3.6 For a broom graph Bp,d ,

P EN I(Bp,d ) = 2 d 3.

Proof Let V (Bp,d ) = {u1 , u2 , , ud , v1 , v2 , , vpd } such that u1 , u2 , , ud is a path

on d vertices and v1 , v2 , , vpd are end vertices that are adjacent to ud . An edge e as shown
in Figure 10 is chosen,

vpd

e v3
u1 u2 u3 ud1 ud v2
v1
Figure 10 Bp,d

and e is deleted, we get p d + 1 components, namely (p d) isolated vertices and a path of

order (d 2). By Proposition 2.1, it follows that

P EN I(Bp,d ) = 1 + P EN I(Pd2 ) = 1 + 2 d 4.

Thus

P EN I(Bp,d ) = 2 d 3. 2
Corollary 3.1 For any broom graph, if p d = 2, then

P EN I(Bp,d ) = P EN I(L(Bp,d )).

Definition 3.5([12]) The join of two graphs G1 (V1 , E1 ) and G2 (V2 , E2 ), denoted by G1 + G2
consists of vertex set V = V1 V2 , and edge set E = E1 E2 and all edges joining V1 with V2 .

Theorem 3.7 For a joint graph K2 + Pp ,

p
P EN I(K2 + Pp ) = 2 p + 2 3.
 Pure Edge-Neighbor-Integrity of Graphs 125

Proof Let K2 be a complete graph with vertices u1 , u2 and Pp , a path with vertices
v1 , v2 , , vp . Let G be the graph K2 + Pp . Then, V (G) = {u1 , u2 , v1 , , vp }, |V (G)| = p + 2,
and |E(G)| = 3p.
The graph K2 + Pp is shown in Figure 11.

u1 u2

v1 v2 v3 v4 vp1 vp

Figure 11 K2 + Pp

Consider 1 = {(u1 , u2 )}, |1 | = 1. Then, G/1 = Pp , so that e (G/1 ) = p 1. Let

2 = {ek = (vk , vk+1 ), 1 k p 1/ek P EN I set of Pp }. Take E1 = {ek /ek P EN I
set of Pp } so that |2 | = |E1 |. Consider = 1 2 . Thus, || = |1 | + |2 | = |1 | + |E1 |
and G/ = Pp /E1 . So e (G/) = e (Pp /E1 ). By Proposition 2.1, we have

|| + e (G/) = |1 | + |E1 | + e (Pp /E1 )

p
= |1 | + P EN I(Pp ) = 1 + 2 p + 2 4 (15)
p
= 2 p + 2 3.

To claim that || + e (G/) is minimum. Suppose 3 is any edge set of G such that
3 = 1 {e} and |3 | = 2. Then |3 | + e (G/3 ) || + e (G/). Let 5 be edge set of G
such that 5 = 2 . Then, e (G/5 ) p. Hence, |5 |+ e (G/5 ) |2 |+ p > ||+ e (G/).
Therefore, from the above discussion, it follows that || + e (G/) is minimum. Hence, from
equation (15) and the minimality of || + e (G/), we have

2
p
P EN I(K2 + Pp ) = 2 p + 2 3.

Theorem 3.8 For a joint graph K2 + Cp ,

p 1, p=3, 4;
P EN I(K2 + Cp ) =
2p 2, p 5.

Proof The proof is similar to that of the Theorem 3.7. 2

Theorem 3.9 For a joint graph K2 + Kp ,

p+2
P EN I(K2 + Kp ) = .
2
126 Sultan Senan Mahde and Veena Mathad

Proof Since K2 + Kp = Kp+2 is a complete graph of order p + 2, by Proposition 2.1,

p+2
P EN I(K2 + Kp ) = P EN I(K(p+2) ) =
2
. 2

References

[1] S. Arumugam and S. Velammal, Edge domination in graphs, Taiwanese Journal of Math-
ematics, 2 (1998), 173-179.
[2] K. S. Bagga, L. W. Beineke, Wayne Goddard, M. J. Lipman and R. E. Pippert, A survey
of integrity, Discrete Applied Math., 37/38 (1992), 13-28.
[3] K. S. Bagga and J. S. Deogun, A variation on the edge integrity, Congress. Numer.,
91(1992), 207-211.
[4] C. A. Barefoot, R. Entringer and H. Swart, Vulnerability in graphs - A comparative survey,
J. Combin. Math. Combin. Comput., 1 (1987), 12-22.
[5] C. A. Barefoot, R. Entringer and H. Swart, Integrity of trees and powers of cycles, Con-
gressus Numerantium, 58 (1987), 103-114.
[6] J. A. Bondy and U. S. R. Murty, Graph Theory, Springer, Berlin, 2008.
[7] M. B. Cozzens and S. Y. Wu, Edge-neighbor-integrity of trees, Australas. J. Combin., 10
(1994), 163-174.
[8] M. B. Cozzens and S. Y. Wu, Vertex neighbor-integrity of trees, Ars Combinatoria, 43
(1996), 169-180.
[9] P. Dundar, The neighbor-integrity of boolean graphs and its compounds, Intern. J. Com-
puter Math., 72 (1999), 441-447.
[10] P. Dundar and A. Ozan, On the neighbor-integrity of sequential joined graphs, Intern. J.
Computer Math., 74 (2000), 45-52.
[11] W. Goddard and H. C. Swart, On the integrity of combinations of graphs, J. Combin.
Math. Combin. Comput., 4 (1988), 3 -18.
[12] F. Harary, Graph Theory, Addison Wesley, Reading Mass, 1969.
[13] D. A. Holton, D. Lou and K. L. Mcavaney, n-Extendability of line graphs, power graphs,
and total graphs, Australasian Journal of Combinatorics, 11 (1995), 215-222.
[14] A. Kirlangic, On the weak-integrity of graphs, Journal of Mathematical Modelling and
Algorithms, 2 (2003), 81-95.
[15] V. Mathad, S. S. Mahde and A. M. Sahal, Vulnerability: vertex neighbor integrity of
middle graphs, Journal of Computer and Mathematical Sciences, 6 (2015), 43-48.
[16] S. Mitchell and S. T. Hedetniemi, Edge domination in trees, Congr. Numer., 19 (1977),
489-509.
[17] M. J. Morgan, S. Mukwembi and H. C. Swart, On the eccentric connectivity index of a
graph, Discrete Math., 311 (2011), 1229-1234.
[18] K. T. Newport and P. K. Varshney, Design of survivable communication networks under
performance constraints, IEEE Transactions on Reliability, 40 (1991), 433-440.
Math.Combin.Book Ser. Vol.1(2017), 127-134

Bounds for the

Largest Color Eigenvalue and the Color Energy

M.A.Sriraj
(Department of Mathematics, Vidyavardhaka College of Engineering, Mysuru - 570002, India)

E-mail: [email protected]

Abstract: In [1] Chandrashekar Adiga et al. introduced the matrix of a vertex colored
graph and studied their eigenvalues called color eigenvalues. Further, defined the color energy
of the graph and obtained some results. In this paper, we obtain bounds for the largest color
eigenvalue and the color energy.

Key Words: Smarandachely vertex coloring, color eigenvalues, color spectral radius, color
energy of a graph.

AMS(2010): 05C15, 05C50

1. Introduction

Let G = (V, E) be finite simple graph with n vertices and m edges. The adjacency matrix
of G is the n n matrix A = A(G), whose entries aij are given by aij = 1 if vi and vj are
adjacent, aij = 0 otherwise. The eigenvalues of A(G) are the eigenvalues of G. The energy
E(G) of a graph G is the sum of the absolute values of the eigenvalues of A(G) [6]. A survey
of development of energy of a graph before 2001 can be found in [7].
Let H be a subgraph of graph G. A Smarandachely vertex coloring respect to H of a graph
G by colors in C is a mapping H : C E(G) such that H (e1 ) 6= H (e2 ) if e1 and e2 are
edges of a subgraph isomorphic to H in G. Particularly, if H = G, such a Smarandachely vertex
coloring is the usual vertex coloring of a graph G, i.e., a coloring of its vertices such that no two
adjacent vertices receive the same color. The minimum number of colors needed for coloring of
a graph G is called chromatic number and denoted by (G).
Recently in [1], Chandrashekar Adiga et al. have introduced the n n matrix A = Ac (G)
of a vertex colored graph G, which is defined as follows: If c(vi ) is the color of vi , then

1 if vi and vj are adjacent,


aij = 1 if vi and vj are non-adjacent with c(vi ) = c(vj ),


0 otherwise.

1 Received June 26, 2016, Accepted February 20, 2017.

128 M.A. Sriraj

The eigenvalues of Ac (G) are called color eigenvalues of G. The color energy Ec (G) is
defined to be the sum of the absolute values of the color eigenvalues of G. In [1]C.Adiga et
al. have computed the color energy E (G) of few families of graphs with minimum number
of colors. In [2] they have also derived explicit formulas for the color energies of the unitary
Cayley graph, the complement of the colored unitary Cayley graph and the gcd-graphs.
The main purpose of this paper is to establish some bounds for largest color eigenvalue
and color energy. In literature there are several upper bounds for the spectral radius 1 of a
graph G. For more details see [3], [4], [5] and [8].

2. Bounds for the Largest Color Eigenvalue

First we prove the following theorem which is useful to obtain bounds for the largest color
eigenvalue of a graph G.

Theorem 2.1 Let G be a colored graph with n vertices and m edges and H be a (n, m1 )-graph.
If 1 2 n are color eigenvalues of G and 1 2 n are eigenvalues of
H, then
Xn p
i i 2 (m + mc )m1 ,
i=1

where mc is the number of pairs of non-adjacent vertices receiving the same color in G.

Proof By Cauchy-Schwarz inequality we have

n
!2 n
! n
!
2
X X X
i i 2i i (2.1)
i=1 i=1 i=1

n n
2
X X
In [1], it has been proved that 2i = 2(m + mc ). It is well-known that i = m1 . Using
i=1 i=1
these in (2.1) we obtain
n
!
2
X p
i i 2 (m + mc )m1 .
i=1

If we know the spectrum of a graph H with n vertices and m1 edges, then we can find an
upper bound for the largest color eigenvalue of the colored graph G with n vertices.
Using the above theorem we establish bounds for the largest color eigenvalue.

Proposition 2.2 If G is a colored (n, m)-graph and 1 2 n are color eigenvalues

of G, then
p
" #
1 p X
1 2(m + mc )p(p 1) + np+i ,
p1 i=2

where p is any integer 1 < p n.

 Bounds for the Largest Color Eigenvalue and the Color Energy 129

S
Proof Let H = Kp Knp . Then the spectrum of H is

(p 1) 0 1
.
1 np p1

Then by Theorem 2.1 we have

1 (p 1) + 2 (0) + 3 (0) + + np+1 (0) + np+2 (1) + + n (1)

r
(m + mc )p(p 1)
2 .
2
Thus,
p
X
p
(p 1)1 2(m + mc )p(p 1) + np+i
i=2

Hence,
p
" #
1
2
p X
1
2(m + mc )p(p 1) + np+i .
p1 i=2

Remark 2.3 If p = n in the above proposition, then

r
2(m + mc )(n 1)
1 .
n

Remark 2.4 If p = 2 in the above proposition, then

p
1 n 2 (m + mc ).

Proposition 2.5 If G is a colored (n, m)-graph and 1 2 n are color eigenvalues

of G, then s
k
X 2(m + mc )k(p 1)
i ,
i=1
p
n
where p is any integer 1 p n and k = p.

Proof Let H be a graph with n vertices and k components each is a complete graph Kp .
Then n = pk and H has kp(p1)
2 edges. Thus spectrum of H is

(p 1) 1
.
k k(p 1)

Then by Theorem 2.1 we have

(p 1)1 + (p 1)2 + + (p 1)k + (1)k+1 + + (1)n

r
(m + mc )kp(p 1)
2 .
2
130 M.A. Sriraj

Therefore,
k
X n
X p
p i i 2(m + mc )kp(p 1)
i=1 i=1

and s
k
2(m + mc )k(p 1)
2
X
i .
i=1
p

Remark 2.6 If k = 1 in the above proposition, then

s
2(m + mc )(p 1)
1 .
p

Proposition 2.7 If G is a colored (n, m)-graph and 1 2 n are color eigenvalues

of G, then " k k
#
X X p
i nk+i 2 (m + mc )k,
i=1 i=1

where 1 k < n and k|n.

Proof Let H be a graph with n vertices and k components each is a complete bipartite
graph Kp,q . Then n = k(p + q) and H has kpq edges. Thus, the spectrum of H is

pq 0 pq
.
k k(p + q 2) k

Then, by Theorem 2.1 we have

pq1 + + pqk + (0)k+1 + + (0)k+k(p+q2) + ( pq)k(p+q1)+1 +
p
+( pq)n 2 (m + mc )kpq.

Therefore, " k k
#
X X p
pq i nk+i 2 (m + mc )kpq
i=1 i=1

and " k k
#
2
X X p
i nk+i 2 (m + mc )k.
i=1 i=1

Remark 2.8 If k = 1 in the above proposition, then

p
1 n 2 (m + mc ).

3. Bounds for Color Energy of a Graph

In [1] Adiga et al. have proved the following results.

 Bounds for the Largest Color Eigenvalue and the Color Energy 131

n
X
Proposition 3.1 If 1 , 2 , . . . , n are color eigenvalues of Ac (G), then 2i = 2(m + mc ),
i=1
where mc is the number of pairs of non-adjacent vertices receiving the same color in G.

Theorem 3.2 Let G be a connected colored graph with n vertices, m edges, and mc be number
p
of pairs of non-adjacent vertices receiving the same color. Then Ec (G) 2n(m + mc ).

Using Proposition 3.1 and the Theorem 3.2 we prove the following result.

Theorem 3.3 Let G be a connected colored graph with n vertices and m edges. Then
p p
2 m + mc Ec (G) 2 m(m + mc ).

Proof Let 1 , 2 , . . . , n be the color eigenvalues of G. Since

n
X n
X
i = 0 and i 2 = 2(m + mc )
i=1 i=1

we have X
i j = (m + mc ). (3.1)
i<j

Now consider
n
!2 n n
X X X
2
[Ec (G)] = |i | = |i | |j |
i=1 i=1 j=1
n
X X
= |i |2 + 2 |i ||j |
i=1 1i<jn
n
X X
|i |2 + 2 | i j | 2(m + mc ) + 2(m + mc )
i=1 i<j

by using Proposition 3.1 and equation (3.1). Hence

p
Ec (G) 2 m + mc .
p
From Theorem 3.2, we have Ec (G) 2n(m + mc ). Since n 2m, we have Ec (G)
p
2 m(m + mc ). Thus,

2
p p
2 m + mc Ec (G) 2 m(m + mc ).

4. Bounds for Color Spectral Radius and Color Energy

We now establish a lower bound and an upper bound for color spectral radius. Also using these
132 M.A. Sriraj

bounds we establish bounds for color energy.

Proposition 4.1 Let G be a colored (n, m)-graph and c (G) = max {|i |} be the color spectral
1in
radius of G. Then r
2(m + mc ) p
c (G) 2(m + mc ).
n

Proof Consider

2 c (G) = max {|i |2 }

1in
Xn
j 2 = 2(m + mc ).
j=1

So,
p
c (G) 2(m + mc ).

Next consider
n
X
n 2 c (G) i 2
i=1
2(m + mc ).

we have,
r
2(m + mc )
c (G) .
n
Therefore, r
2(m + mc )
2
p
c (G) 2(m + mc ).
n

Theorem 4.2 Let G be a colored graph and 1 , 2 , . . . , n be the color eigenvalues of G. If

n 2(m + mc ) and 1 2(m+m
n
c )
, then
v "
u 2 #
2(m + mc ) u 2(m + mc )

Ec (G)
+ (n 1) 2(m + mc )
t .
n n

Proof We have
n
X
i 2 = 2(m + mc ) 1 2 . (4.1)
i=2
 Bounds for the Largest Color Eigenvalue and the Color Energy 133

By a special case of Cauchy-Schwarz inequality we have

n
!2 n
X X
|i | n |i |2 .
i=1 i=1

Thus,
n
!2 n
X X
|i | (n 1) |i |2
i=2 i=2

and hence v
n n
! u
X X
|i |2 .
u
|i | t(n 1) (4.2)
i=2 i=2

Employing (4.1) in (4.2), we obtain

q
Ec (G) 1 (n 1)[2(m + mc ) 1 2 ].

i.e,
q
Ec (G) 1 + (n 1)[2(m + mc ) 1 2 ].

Consider the function

p
F (x) = x + (n 1)[2(m + mc ) x2 ].

Then
p
x (n 1)
F (x) = 1 p .
2(m + mc ) x2
q 
2(m+mc )
p
Observe that F (x) is decreasing in ,
2(m + mc ) .
n

2(m+mc )
Since n 2(m + mc ) and n 1 , we have
r
2(m + mc ) 2(m + mc ) p
< 1 2(m + mc ).
n n
Last inequality follows from Proposition 4.1.

Hence
v "
u 2 #
2(m + mc ) u 2(m + mc )

Ec (G)
n
t
+ (n 1) 2(m + mc )
n
. 2

As the proof of the following theorem is similar to that of Theorem 4.2 we omit the proof.
134 M.A. Sriraj

q
2(m+mc ) 2(m+mc )
Theorem 4.3 If n 2(m + mc ) and n c (G) n , then
v "
u

  #
) 2
2(m + mc ) u 2(m + m c
Ec (G) + t(n 1) 2(m + mc ) .
n n

References

[1] C. Adiga, E. Sampathkumar, M.A. Sriraj, Shrikanth A. S, Colored energy of a graph,

Proceedings of Jangeon Math. Society, 16 (3)(2013), 335-351.
[2] C. Adiga, E. Sampathkumar, M.A. Sriraj,Color energy of Unitary Cayley graph,Discussiones
Mathematicae Graph Theory 34 (2014), 707-721.
[3] A.Berman and X.D.Zang, On the spectral radius of graphs with cut vertices, Journal of
Combinatorial Theory, Ser.B, 83 (2001), 233-40.
[4] R.C.Brigham and R.P. Dutton, Bounds on graph spectra, J. Combin. Theory, Ser. B
37(1984), 228-234.
[5] K.C. Das and Pavan Kumar, Bounds on the greatest eigenvalues of graphs, Indian J.Pure
Appl. Math., 34(6) (2003), 917-925.
[6] I.Gutman, The energy of a graph, Ber. Math. Stat. Sekt. Forschungsz. Graz, 103(1978),
1-22.
[7] I.Gutman, The energy of a graph: old and new results, Combinatorics and applications, A.
Betten, A. Khoner, R. Laue and A. Wassermann, eds., Springer, Berlin, 2001, 196-211.
[8] H. S. Ramane and H. B. Walikar, Bounds for the eigenvalues of a graph, Graphs, Com-
binatorics, Algorithms and Applications, (eds.: S. Arumugam, B. D. Acharya and S. B.
Rao), Narosa Publishing House, New Delhi, 2005, 115-122.
Math.Combin.Book Ser. Vol.1(2017), 135-138

A Note on Acyclic Coloring of Sunlet Graph Families

Kowsalya.V
Part-Time Research Scholar (Category-B), Research & Development Centre

Bharathiar University, Coimbatore 641 046, Tamil Nadu, India

Vernold Vivin.J
Department of Mathematics, University College of Engineering Nagercoil

Anna University Constituent College, Konam, Nagercoil-629 004, Tamil Nadu, India

E-mail: [email protected], [email protected]

Abstract: In this paper, we find the acyclic chromatic number a for the central graph of
sunlet graph C(Sn ), line graph of sunlet graph L(Sn ), middle graph of sunlet graph M (Sn )
and the total graph of sunlet graph T (Sn ) for all n > 3.

Key Words: Smarandachely vertex coloring, acyclic coloring, sunlet graph, central graph,
line graph, middle graph and total graph.

AMS(2010): 05C15, 05C75.

1. Introduction

Let G be a finite graph and let H G be a subgraph of G. A Smarandachely vertex coloring

respect to a subgraph H G by colors in C is a mapping H : C E(G) such that H (e1 ) 6=
H (e2 ) if e1 and e2 are edges of a subgraph isomorphic to H in G. Particularly, let H = G.
Then, such a Smarandachely vertex coloring is clearly the usual proper vertex coloring (or
proper coloring) of G, i.e., a coloring : V N + on G such that if v and u are adjacent
vertices, then (v) 6= (u). The chromatic number of a graph G is the minimum number
of colors required in any proper coloring of G. Generally, The notion of acyclic coloring was
introduced by Branko Gr unbaum in 1973. An acyclic coloring of a graph G is a proper vertex
coloring such that the induced subgraph of any two color classes is acyclic, i.e., disjoint collection
of trees. The minimum number of colors needed to acyclically color the vertices of a graph G
is called as acyclic chromatic number and is denoted by a (G).

2. Preliminaries

A sunlet graph on 2n vertices is obtained by attaching n pendant edges to the cycle Cn and
denoted by Sn .
1 Received July 26, 2016, Accepted February 28, 2017.
136 Kowsalya.V and Vernold Vivin.J

For a given graph G = (V, E) we do an operation on G by subdividing each edge exactly

once and joining all the non-adjacent vertices of G. The graph obtained by this process is called
central graph [5] of G denoted by C(G).
A line graph [1, 4] of a graph G, denoted by L(G), is a graph whose vertices are the edges
of G, and if u, v E(G) then uv E(L(G)) if u and v share a vertex in G.
A middle graph [3] of G, is defined with the vertex set V (G) E(G) where two vertices are
adjacent iff they are either adjacent edges of G or one is the vertex and the other is an edge
incident with it and it is denoted by M (G).
The total graph [1 ,3, 4] of G has vertex set V (G) E(G), and edges joining all elements
of this vertex set which are adjacent or incident in G.
Additional graph theory terminology used in this paper can be found in [1, 4].
In the following sections we find the acyclic chromatic number for the central graph of
sunlet graph C(Sn ), line graph of sunlet graph L(Sn ), middle graph of sunlet graph M (Sn ) and
the total graph of sunlet graph T (Sn ).

Definition 2.1([2]) An ayclic coloring of a graph G is a proper coloring such that the union
of any two color classes induces a forest.

3. Acyclic Coloring on Central Graph of Sunlet Graph

Theorem 3.1 Let Sn be a sunlet graph with 2n vertices, then

a (C(Sn )) = n, n 3.

S S
Proof Let V (Sn ) = {u1 , u2 , , un } {v1 , v2 , , vn } and E(Sn ) = {ei : 1 i n}
{ei : 1 i n}, where ei is the edge vi vi+1 (1 i n 1), en is the edge vn v1 and ei is the
edge vi ui (1 i n). For 1 i n, ui is the pendant vertex and vi is the adjacent vertex
S S
to ui . By the definition of central graph V (C(Sn )) = V (Sn ) E(Sn ) = {ui : 1 i n}
S
{vi : 1 i n} {vi : 1 i n} {ui : 1 i n}, where vi and ui represents the edge ei
and ei , (1 i n) respectively.
Assign the following coloring for C(Sn ) as acyclic:

(1) For 1 i n assign the color ci to ui , vi ;

(2) For 1 i n 1 assign the color ci+1 to ui and c1 to un ;
(3) For 1 i n 1 assign the color ci to vi and cn to v1 .
Thus, a (C(Sn )) = n, for if a (C(Sn )) < n, say n 1. A contradiction to proper coloring since,
n, {ui : 1 i n} forms a clique of order n. Hence, a (C(Sn )) = n, n 3. 2

4. Acyclic Coloring on Line Graph of Sunlet Graph

Theorem 4.1 Let n 3 be a positive integer, then a (L(Sn )) = 3.

 A Note on Acyclic Coloring of Sunlet Graph Families 137

S
Proof Let V (Sn ) = {u1 , u2 , , un } {v1 , v2 , , vn } and E(Sn ) = {ei : 1 i n}
S
{ei : 1 i n}, where ei is the edge vi vi+1 (1 i n 1), en is the edge vn v1 and

ei is the edge vi ui (1 i n). By the definition of line graph V (L(Sn )) = E(Sn ) =
{ui : 1 i n} {vi : 1 i n 1} {vn } where vi and ui represents the edge ei and ei ,
S

(1 i n) respectively.
Assign the coloring as acyclic as follows:

(1) For 1 i 3, assign the vertices vi as (vi ) = ci , and for 4 i n, let (vi ) = c1 if
i 1 mod 3, (vi ) = c2 if i 2 mod 3, (vi ) = c3 if i 0 mod 3;
(2) For 1 i n, assign the vertices ui with colors c1 , c2 , c3 such that (ui ) 6= (vi1

)

and (ui ) 6= (vi ), where v0 = vn .
Thus, a (L(Sn )) = 3, n 3.
To the contrary, let a (L(Sn )) < 3, say 2. A contradiction to proper coloring, since, for
{1 i n 1}, vi , ui+1 , vi+1

2

is a complete graph K3 . Hence, a (L(Sn )) = 3, n 3.

5. Acyclic Coloring on Middle Graph of Sunlet Graph

Theorem 5.1 Let n 3 be a positive integer, then a (M (Sn )) = 4.

Proof Let V (Sn ) = {u1 , u2 , , un } {v1 , v2 , , vn } and E(Sn ) = {ei : 1 i n} {ei :

S S

1 i n}, where ei is the edge vi vi+1 (1 i n 1), en is the edge vn v1 and ei is

S
the edge vi ui (1 i n). By definition of middle graph V (M (Sn )) = V (Sn ) E(Sn ) =
{ui : 1 i n} {vi : 1 i n} {vi : 1 i n} {ui : 1 i n}, where vi and ui repre-
S S S

sents the edge ei and ei , (1 i n) respectively.

Define the mapping such that (V (M (Sn ))) ci for 1 i 4 as follows:

(1) For 1 i n, assign the vertices vi as

(vi ) = {c1 c2 c3 c1 c2 c3 c1 c2 c3 } if n 0 mod 3,

(vi ) = {c1 c2 c3 c1 c2 c3 c1 c2 c3 c2 } if n 1 mod 3,
(vi ) = {c1 c2 c3 c1 c2 c3 c1 c2 c3 c1 c2 } if n 2 mod 3;
(2) Assign (ui ) = (vi ) = c4 for 1 i n;
(3) Assign the vertices ui with c1 , c2 , c3 such that (ui ) 6= (vi1

) and (ui ) 6= (vi ) for
1 i n where v0 = vn .
Thus, a (M (Sn )) = 4, n 3.
To the contrary, let a (M (Sn )) < 4, say 3. A contradiction to proper coloring, since n,

vi1 , vi , vi , ui , where v0 = vn , forms a clique of order 4. Thus is a proper acyclic coloring

and hence a (M (Sn )) = 3, n 3. 2

6. Acyclic Coloring on Total Graph of Sunlet Graph

Theorem 6.1 Let Sn be a sunlet graph with 2n vertices then for n 3, a (T (Sn )) = 6.
138 Kowsalya.V and Vernold Vivin.J

Proof Let V (Sn ) = {u1 , u2 , , un } {v1 , v2 , , vn } and E(Sn ) = {ei : 1 i n} {ei :

S S

1 i n}, where ei is the edge vi vi+1 (1 i n 1), en is the edge vn v1 and ei is

S
the edge vi ui (1 i n). By the definition of total graph V (T (Sn )) = V (Sn ) E(Sn ) =
{ui : 1 i n} {vi : 1 i n} {vi : 1 i n} {ui : 1 i n}, where vi and ui repre-
S S S

sents the edge ei and ei , (1 i n) respectively.

Define the mapping such that (V (T (Sn ))) ci for 1 i 6 as follows:

(1) For 1 i 3, let (vi ) = ci and for 4 i n, let

(vi ) = c1 if i 1 mod 3,
(vi ) = c2 if i 2 mod 3,
(vi ) = c3 if i 0 mod 3;

(2) For 1 i 3, let (vi ) = ci+3 and for 4 i n, let

(vi ) = c4 if i 1 mod 3,
(vi ) = c5 if i 2 mod 3,
(vi ) = c6 if i 0 mod 3;
(3) Let (ui ) = (vi ) + 1 for 1 i n;
(4) For 1 i n, let (ui ) = (vi ) + 1 and (ui ) = c1 if (vi ) = c6 .
Thus, a (T (Sn )) = 6 for n 3.
For 1 i 6, the union of any two color classes ci1 and ci induces subgraph whose
components are trees, hence by Definition 1.1, is a proper acyclic coloring and a (T (Sn )) = 6
for n 3. 2

References

[1] Bondy.J.A and Murty.U.S.R, Graph theory with Applications, London, MacMillan 1976.
[2] Branko Gr unbaum, Acyclic colorings of planar graphs, Israel J.Math., 14(1973), 390408.
[3] Danuta Michalak, On middle and total graphs with coarseness number equal 1, Springer
Verlag Graph Theory, Lagow Proceedings, Berlin Heidelberg, New York, Tokyo, (1981),
139150.
[4] Frank Harary, Graph Theory, Narosa Publishing home, New Delhi 1969.
[5] Vernold Vivin.J, Harmonious coloring of total graphs, n-leaf, central graphs and circumde-
tic graphs, Ph.D thesis, Bharathiar University, 2007, Coimbatore, India.
Famous Words 139

In almost every face and every person, they may discover fine feathers and
defects, good and bad qualities.

By Benjamin Franklin, an American polymath and a leading author, printer, political

theorist, politician, freemason, postmaster, scientist, inventor, civic activist, statesman, and
diplomat.
Author Information

Submission: Papers only in electronic form are considered for possible publication. Papers
prepared in formats, viz., .tex, .dvi, .pdf, or.ps may be submitted electronically to one member
of the Editorial Board for consideration both in International Journal of Mathematical
Combinatorics and Mathematical Combinatorics (International Book Series). An
effort is made to publish a paper duly recommended by a referee within a period of 3 4
months. Articles received are immediately put the referees/members of the Editorial Board
for their opinion who generally pass on the same in six weeks time or less. In case of clear
recommendation for publication, the paper is accommodated in an issue to appear next. Each
submitted paper is not returned, hence we advise the authors to keep a copy of their submitted
papers for further processing.

Abstract: Authors are requested to provide an abstract of not more than 250 words, lat-
est Mathematics Subject Classification of the American Mathematical Society, Keywords and
phrases. Statements of Lemmas, Propositions and Theorems should be set in italics and ref-
erences should be arranged in alphabetical order by the surname of the first author in the
following style:

Books
[4]Linfan Mao, Combinatorial Geometry with Applications to Field Theory, InfoQuest Press,
2009.
[12]W.S.Massey, Algebraic topology: an introduction, Springer-Verlag, New York 1977.

Research papers

[6]Linfan Mao, Mathematics on non-mathematics - A combinatorial contribution, International

J.Math. Combin., Vol.3(2014), 1-34.
[9]Kavita Srivastava, On singular H-closed extensions, Proc. Amer. Math. Soc. (to appear).

Figures: Figures should be drawn by TEXCAD in text directly, or as EPS file. In addition,
all figures and tables should be numbered and the appropriate space reserved in the text, with
the insertion point clearly indicated.

Copyright: It is assumed that the submitted manuscript has not been published and will
not be simultaneously submitted or published elsewhere. By submitting a manuscript, the
authors agree that the copyright for their articles is transferred to the publisher, if and when,
the paper is accepted for publication. The publisher cannot take the responsibility of any loss
of manuscript. Therefore, authors are requested to maintain a copy at their end.

Proofs: One set of galley proofs of a paper will be sent to the author submitting the paper,
unless requested otherwise, without the original manuscript, for corrections after the paper is
accepted for publication on the basis of the recommendation of referees. Corrections should be
restricted to typesetting errors. Authors are advised to check their proofs very carefully before
return.
 March 2017

Contents
Special Smarandache Curves According to Bishop Frame in Euclidean
Spacetime By E. M. Solouma and M. M. Wageeda . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 01
Spectra and Energy of Signed Graphs By Nutan G. Nayak. . . . . . . . . . . . . . . . . . . . . . .10
On Transformation and Summation Formulas for Some Basic
Hypergeometric Series By D.D.Somashekara, S.L.Shalini and K.N.Vidya . . . . . . . . . . . 22
Some New Generalizations of the Lucas Sequence
By F
ugen TORUNBALCI AYDIN and Salim YUCE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
Fixed Point Results Under Generalized Contraction Involving
Rational Expression in Complex Valued Metric Spaces By G. S. Saluja . . . . . . . . 53
A Study on Cayley Graphs over Dihedral Groups By A.Riyas and K.Geetha . . . 63
On the Second Order Mannheim Partner Curve in E 3
By S
eyda Klcoglu and S
uleyman Senyurt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
The -Change of Special Finsler Spaces
By H.S.Shukla, O.P.Pandey and Khageshwar Manda . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
Peripheral Distance Energy of Graphs
By Kishori P. Narayankar and Lokesh S. B. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
Some Properties of a h-Randers Finsler Space
By V.K.Chaubey, Arunima Mishra and A.K.Pandey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
Pure Edge-Neighbor-Integrity of Graphs
By Sultan Senan Mahde and Veena Mathad . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
Bounds for the Largest Color Eigenvalue and the Color Energy
By M.A.Sriraj . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
A Note on Acyclic Coloring of Sunlet Graph Families
By Kowsalya.V and Vernold Vivin.J . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135

An International Book Series on Mathematical Combinatorics