ISBN 978-1-59973-120-9

VOLUME 1, 2010

MATHEMATICAL COMBINATORICS
(INTERNATIONAL BOOK SERIES)

Edited By Linfan MAO

THE MADIS OF CHINESE ACADEMY OF SCIENCES

April, 2010
Vol.1, 2010 ISBN 978-1-59973-120-9

Mathematical Combinatorics
(International Book Series)
Edited By Linfan MAO

The Madis of Chinese Academy of Sciences

April, 2010
Aims and Scope: The Mathematical Combinatorics (International Book Series)
(ISBN 978-1-59973-120-9) is a fully refereed international book series, and published quarterly
comprising 100-150 pages approx. per volume, which publishes original research papers and
survey articles in all aspects of Smarandache multi-spaces, Smarandache geometries, mathe-
matical 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;
Differential Geometry; Geometry on manifolds;
Topological graphs; Algebraic graphs; Random graphs; Combinatorial maps; Graph and
map enumeration; Combinatorial designs; Combinatorial enumeration;
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 fur Mathematik(Germany),

Referativnyi Zhurnal (Russia), Mathematika (Russia), Computing Review (USA), Institute for
Scientific Information (PA, USA), Library of Congress Subject Headings (USA).

Subscription A subscription can be ordered by a mail or 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 H.Iseri
Mansfield University, USA
Email: [email protected]
Editor-in-Chief
Linfan MAO M.Khoshnevisan
Chinese Academy of Mathematics and System School of Accounting and Finance,
Griffith University, Australia
Science, P.R.China
Email: [email protected] Xueliang Li
Nankai University, P.R.China
Email: [email protected]
Editors
Han Ren
East China Normal University, P.R.China
S.Bhattacharya
Email: [email protected]
Deakin University
Geelong Campus at Waurn Ponds W.B.Vasantha Kandasamy
Australia Indian Institute of Technology, India
Email: [email protected] Email: [email protected]

Mingyao Xu
An Chang
Peking University, P.R.China
Fuzhou University, P.R.China
Email: [email protected]
Email: [email protected]
Guiying Yan
Junliang Cai
Chinese Academy of Mathematics and System
Beijing Normal University, P.R.China
Science, P.R.China
Email: [email protected]
Email: [email protected]
Yanxun Chang Y. Zhang
Beijing Jiaotong University, P.R.China Department of Computer Science
Email: [email protected] Georgia State University, Atlanta, USA
Shaofei Du
Capital Normal University, P.R.China
Email: [email protected]

Florentin Popescu and Marian Popescu

University of Craiova
Craiova, Romania

Xiaodong Hu
Chinese Academy of Mathematics and System
Science, P.R.China
Email: [email protected]

Yuanqiu Huang
Hunan Normal University, P.R.China
Email: [email protected]
ii International Journal of Mathematical Combinatorics

If the facts don’t fit the theory, change the facts.

By Albert Einstein, an American theoretical physicist.

Math.Combin.Book Ser. Vol.1 (2010), 01-04

Singed Total Domatic Number of a Graph

H.B. Walikar1 , Shailaja S. Shirkol2 , Kishori P.Narayankar3

1. Department of Computer Science, Karnatak University, Dharwad, Karnataka, INDIA-580003

2. Department of Mathematics, Karnatak University, Dharwad, Karnataka, INDIA

3. Department of Mathematics, Mangalore University, Mangalore, Karnataka, INDIA

Email: [email protected], shaila [email protected]

Abstract: Let G be a finite and simple graph with vertex set V (G), k ≥ 1 an integer and
P
let f : V (G) → {−k, k − 1, · · · , −1, 1, · · · , k − 1, k} be 2k valued function. If f (x) ≥ k
x∈N(v)
for each v ∈ V (G), where N (v) is the open neighborhood of v, then f is a Smarandachely
k-Signed total dominating function on G. A set {f1 , f2 , . . . , fd } of Smarandachely k-Signed
P d
total dominating function on G with the property that fi (x) ≤ k for each x ∈ V (G) is
i=1
called a Smarandachely k-Signed total dominating family (function) on G. Particularly, a
Smarandachely 1-Signed total dominating function or family is called signed total dominating
function or family on G. The maximum number of functions in a signed total dominating
family on G is the signed total domatic number of G. In this paper, some properties related
signed total domatic number and signed total domination number of a graph are studied
and found the sign total domatic number of certain class of graphs such as fans, wheels and
generalized Petersen graph.

Key Words: Smarandachely k-signed total dominating function, signed total domination
number, signed total domatic number.

AMS(2000): 05C69

§1. Terminology and Introduction

Various numerical invariants of graphs concerning domination were introduced by means of

dominating functions and their variants [1] and [4]. We considered finite, undirected, simple
graphs G = (V, E) with vertex set V (G) and edge set E(G). The order of G is given by
n = |V (G)|. If v ∈ V (G), then the open neighborhood of v is N (v) = {u ∈ V (G)|uv ∈ E(G)}
and the closed neighborhood of v is N [v] = {v} ∪ N (v). The number dG (v) = d(v) = |N (v)|
is the degree of the vertex v ∈ V (G), and δ(G) is the minimum degree of G. The complete
graph and the cycle of order n are denoted by Kn and Cn respectively. A fan and a wheel
is a graph obtained from a path and a cycle by adding a new vertex and edges joining it to
all the vertices of the path and cycle respectively. The generalized Petersen graph P (n, k) is
defined to be a graph on 2n vertices with V (P (n, k)) = {vi ui : 1 ≤ i ≤ n} and E(P (n, k)) =
1 Received February 18, 2010. Accepted March 20, 2010.
2 H.B. Walikar, Shailaja S. Shirkol, Kishori P.Narayankar

{vi vi+1 , vi ui , ui ui+k : 1 ≤ i ≤ n, subscripts modulo n}. If A ⊆ V (G) and f is a mapping from
P
V (G) in to some set of numbers, then f (A) = x∈A f (x).
Let k ≥ 1 be an integer and let f : V (G) → {−k, k − 1, · · · , −1, 1, · · · , k − 1, k} be 2k
P
valued function. If f (x) ≥ k for each v ∈ V (G), where N (v) is the open neighborhood of
x∈N (v)
v, then f is a Smarandachely k-Signed total dominating function on G. A set {f1 , f2 , . . . , fd } of
Pd
Smarandachely k-Signed total dominating function on G with the property that fi (x) ≤ k
i=1
for each x ∈ V (G) is called a Smarandachely k-Signed total dominating family (function) on
G. Particularly, a Smarandachely 1-Signed total dominating function or family is called signed
total dominating function or family on G. The singed total dominating function is defined in [6]
P
as a two valued function f : V (G) → {−1, 1} such that x∈N (v) f (x) ≥ 1 for each v ∈ V (G).
The minimum of weights w(f ), taken over all signed total dominating functions f on G, is
called the signed total domination number γts (G). Signed total domination has been studied in
[3].
A set {f1 , f2 , . . . , fd } of signed total dominating functions on G with the property that
Pd
i=1 fi (x) ≤ 1 for each x ∈ V (G), is called a signed total dominating family on G. The
maximum number of functions in a signed total dominating family is the signed total domatic
number of G, denoted by dst (G). Signed total domatic number was introduced by Guan Mei
and Shan Er-fang [2]. Guan Mei and Shan Er-fang [2] have determined the basic properties of
dst (G). Some of them are analogous to those of the signed domatic number in [5] and studied
sharp bounds of the signed total domatic number of regular graphs, complete bipartite graphs
and complete graphs. Guan Mei and Shan Er-fang [2] presented the following results which are
useful in our investigations.

Proposition 1.1([6]) For Circuit Cn of length n we have γts (Cn ) = n.

Proof Here no other signed total dominating exists than the constants equal to 1. 

Theorem 1.2([3]) Let T be a tree of order n ≥ 2. then, γts (T ) = n if and only if every vertex
of T is a support vertex or is adjacent to a vertex of degree 2.

Proposition 1.3([2]) The signed total domatic number dst (G) is well defined for each graph G.

Proposition 1.4([2]) For any graph G of order n ,γts (G) · dst (G) ≤ n.

Proposition 1.5([2]) If G is a graph with the minimum degree δ(G), then 1 ≤ dst (G) ≤ δ(G).

Proposition 1.6([2]) The signed total domatic number is an odd integer.

Corollary 1.7([2]) If G is a graph with the minimum degree δ(G) = 1 or 2, then dst (G) = 1.
In particular, dst (Cn ) = dst (Pn ) = dst (K1,n−1 ) = dst (T ) = 1, where T is a tree.

§2. Properties of the Signed Total Domatic Number

Proposition 2.1 If G is a graph of order n and γts (G) ≥ 0 then, γts (G)+ dst (G) ≤ n+ 1 equality
 Singed Total Domatic Number of a Graph 3

holds if and only if G is isomorphic to Cn or tree T of order n ≥ 2.

Proof Let G be a graph of order n. The inequality follows from the fact that for any two
non-negative integers a and b, a + b ≤ ab + 1. By Proposition 1.4 we have,

γts (G) + dst (G) ≤ γts (G) · dst (G) + 1 ≤ n + 1

Suppose that γts (G) + dst (G) = n + 1 then, n + 1 = γts (G) + dst (G) ≤ γts (G) · dst (G) + 1 ≤ n + 1.
This implies that γts (G) + dst (G) = γts (G) · dst (G) + 1. This shows that γts (G) · dst (G) = n
Solving equations 1 and 2 simultaneously, we have either γts (G) = 1 and dst (G) = n or γts (G) = n
and dst (G) = 1. If γts (G) = 1 and dst (G) = n then n = dst (G) ≤ δ(G) There fore, δ(G) ≥ n a
contradiction.
If γts (G) = n and dst (G) = 1 then by Proposition 1.1 and Proposition 1.2, we have γts (Cn ) =
n and dst (Cn ) = 1 and By Theorem 1.2, If T is a tree of order n ≥ 2 then, γts (T ) = n if and only
if every vertex of T is a support vertex or is adjacent to a vertex of degree 2 and dst (T ) = 1. 

Theorem 2.2 Let G be a graph of order n then dst (G) + dst (Ḡ) ≤ n − 1.

Proof Let G be a regular graph order n, By Proposition 1.5 we have dst (G) ≤ δ(G) and
dst (Ḡ) ≤ δ(Ḡ). Thus we have,

dst (G) + dst (Ḡ) ≤ δ(G) + δ(Ḡ) = δ(G) + (n − 1 − ∆(G)) ≤ n − 1.

Thus the inequality holds. 

§3. Signed Total Domatic Number of Fans, Wheels and Generalized

Petersen Graph

Proposition 3.1 Let G be a fan of order n then dst (G) = 1.

Proof Let n ≥ 2 and let x1 , x2 , . . . , xn be the vertex set of the fan G such that x1 , x2 , . . . , xn , x1
is a cycle of length n and xn is adjacent to xi for each i = 2, 3, . . . , n − 2. By Proposition 1.5
and Proposition 1.6, 1 ≤ dst (G) ≤ δ(G) = 2, which implies dst (G) = 1 which proves the result.

Proposition 3.2 If G is a wheel of order n then dst (G) = 1.

Proof Let x1 , x2 , . . . , xn be the vertex set of the wheel G such that x1 , x2 , . . . , xn−1 , x1 is
a cycle of length n − 1 and xn is adjacent to xi for each i = 1, 2, 3, . . . , n − 1. According to the
Proposition 1.5 and Proposition 1.6, we observe that either dst (G) = 1 or dst (G) = 3. Suppose
to the contrary that dst (G) = 3. Let {f1 , f2 , f3 } be a corresponding signed total dominating
family. Because of f1 (xn ) + f2 (xn ) + f3 (xn ) ≤ 1, there exists at least one function say f1 with
P
f1 (xn ) = −1 The condition x∈N (v) f1 (x) ≥ 1 for each v ∈ (V (G) − {xn }) yields f1 (x) = 1 for
each some i ∈ {1, 2, . . . , n − 1} and t = 2, 3 then it follows that ft (xi+1 ) = ft (xi+2 ) = 1, where
the indices are taken taken modulo n − 1 and ft (xn ) = 1. Consequently, the function ft has
 
at most n2 − 1 for n is odd and n2 − 1 for n is even number of vertices x ∈ V (G) such that
4 H.B. Walikar, Shailaja S. Shirkol, Kishori P.Narayankar

 
ft (x) = −1. Thus there exist at most n2 − 1 for n is odd and n2 − 1 for n is even number of
vertices x ∈ V (G) such that ft (x) = −1 for at least one i = 1, 2, 3. Since n ≥ 4, we observe



that 2(⌊ n2 ⌋ + 1) = 2 n2 − 1 + 1 < n for n is odd and 2 n2 − 1 + 1 < n, a contradiction to
f1 (xn ) + f2 (xn ) + f3 (xn ) ≤ 1 for each x ∈ V (G). 

Proposition 3.3 Let G = P (n, k) be a generalized Petersen graph then for k = 1, 2, dst (G) = 1.

Proof The generalized Petersen graph P (n, 1) is a graph on 2n vertices with

V (P (n, k)) = {vi ui : 1 ≤ i ≤ n}

and E(P (n, k)) = {vi vi+1 , vi ui , ui ui+1 : 1 ≤ i ≤ n, subscripts modulo n}. According to the
Proposition 1.5, Proposition 1.6, we observe that dst (G) = 1 or dst (G) = 3.

Case 1: k=1

Let {f1 , f2 , f3 } be a corresponding signed total dominating functions. Because of f1 (vn ) +

f2 (vn ) + f3 (vn ) ≤ 1 for each i ∈ {1, 2, . . . , 2n}, there exist at least one number j ∈ {1, 2, 3}
such that fj (vi ) = −1. Let, for example, f1 (vk ) = −1 for for any t ∈ {1, 2, . . . , 2n} then
P ∼
x∈N (vt ) f1 (v) ≥ 1 implies that f1 (vk ) = f1 (vk+1 ) = −1 for k = 0, 1mod4 and f1 (vk ) = −1 for
∼
k = 0mod3. This implies, there exist at most 8r, 8r + 2, 8r + 4, 8r + 6, r ≥ 1 vertices such
that ft (v) = −1 for each t = 2, 3 when P (n, 1) is of order 2(6r + l) for 0 ≤ l ≤ 2, 2(6r + 3),
n
2(6r + 4), 2(6r + 5) respectively. Thus there exist 3(8r) = 3(8( 12 − 6l ) < n (similarly < n for
all values of vertex set) a contradiction to f1 (vn ) + f2 (vn ) + f3 (vn ) ≤ 1 for each v ∈ V (G).
Case 2: k=2

Similar to the proof of Case 1, we can prove the claim in this case. 

References

[1] Cockayne E.J, Hedetniemi S.T. Towords a theory of domination in graphs, J. Networks,
1977, 7: 247-261.
[2] Guan Mei, Shan Er-fang. Signed total domatic number of a graph, J.Shanghai Univ (Engl
Ed) 2008 12 (1): 31-34.
[3] Henning M.A. Signed total domination in graphs, J. Discrete Mathematics, 2001, 278:
109-125.
[4] Haynes T.W, Hedetniemi S.T, Slater P.J. Fundamentals of domination in graphs, M. New
York: Marcel Dekker, 1998.
[5] Volkmann L, Zelinka B. Signed domatic number of a graph, J. Discrete Applied Mathe-
matics, 2005, 150: 261-267.
[6] Zelinka B, Liberec. Signed total domination number of a graph, J. Czechoslovak Mathe-
matical Journal, 2001, 51: 225-229.
Math.Combin.Book Ser. Vol.1 (2010), 05-29

Radio Number of Cube of a Path

B. Sooryanarayana1, Vishu Kumar M.2 and Manjula K.3

1. Dept.of Mathematical & Computational Studies, Dr.Ambedkar Institute of Technology,

Bangalore, Karnataka State, Pin 560 056, INDIA

2. Dept.of Mathematics, Reva Institute of Technology, Yelahanka,

Bangalore, Karnataka State, Pin 560 064, INDIA

3. Dept.of Mathematics, Bangalore Institute of Technology, V. V. Puram, Bangalore,

Karnataka State,Pin 560 078, INDIA

Email: dr [email protected]; m [email protected]; manju [email protected]

Abstract: Let G be a connected graph. For any two vertices u and v, let d(u, v) denotes
the distance between u and v in G. The maximum distance between any pair of vertices is
called the diameter of G and is denoted by diam(G). A Smarandachely k-radio labeling of
a connected graph G is an assignment of distinct positive integers to the vertices of G, with
x ∈ V (G) labeled f (x), such that d(u, v) + |f (u) − f (v)| ≥ k + diam(G). Particularly, if
k = 1, such a Smarandachely radio k-labeling is called radio labeling for abbreviation. The
radio number rn(f ) of a radio labeling f of G is the maximum label assignment to a vertex
of G. The radio number rn(G) of G is minimum {rn(f )} over all radio labelings of G. In
this paper, we completely determine the radio number of the graph Pn3 for all n ≥ 4.

Keywords: Smarandachely radio k-labeling, radio labeling, radio number of a graph.

AMS(2010): 05C78, 05C12, 05C15

§1. Introduction

All the graphs considered here are undirected, finite, connected and simple. The length of a
shortest path between two vertices u and v in a graph G is called the distance between u and v
and is denoted by dG (u, v) or simply d(u, v). We use the standard terminology, the terms not
defined here may be found in [1].
The eccentricity of a vertex v of a graph G is the distance from the vertex v to a farthest
vertex in G. The minimum eccentricity of a vertex in G is the radius of G, denoted by r(G),
and the of maximum eccentricity of a vertex of G is called the diameter of G, denoted by
diam(G). A vertex v of G whose eccentricity is equal to the radius of G is a central vertex.
For any real number x, ⌈x⌉ denotes the smallest integer greater than or equal to x and ⌊x⌋
1 Received January 5, 2010. Accepted March 20, 2010.
6 B. Sooryanarayana, Vishu Kumar M. and Manjula K.

denotes the greatest integer less than or equal to x. We recall that k th power of a graph G,
denoted by Gk is the graph on the vertices of G with two vertices u and v are adjacent in Gk
whenever d(u, v) ≤ k. The graph G3 is called a cube of G.
A labeling of a connected graph is an injection f : V (G) → Z + , while a Smarandache k-
radio labeling of a connected graph G is an assignment of distinct positive integers to the vertices
of G, with x ∈ V (G) labeled f (x), such that d(u, v) + |f (u) − f (v)| ≥ k + diam(G). Particularly,
if k = 1, such a Smarandache radio k-labeling is called radio labeling for abbreviation. The
radio number rn(f ) of a radio labeling f of G is the maximum label assigned to a vertex of G.
The radio number rn(G) of G is min{rn(f )}, over all radio labelings f of G. A radio labeling
f of G is a minimal radio labeling of G if rn(f ) = rn(G).
Radio labeling is motivated by the channel assignment problem introduced by Hale et al
[10] in 1980. The radio labeling of a graph is most useful in FM radio channel restrictions to
overcome from the effect of noise. This problem turns out to find the minimum of maximum
frequencies of all the radio stations considered under the network.
The notion of radio labeling was introduced in 2001, by G. Chartrand, David Erwin, Ping
Zhang and F. Harary in [2]. In [2] authors showed that if G is a connected graph of order
n and diameter two, then n ≤ rn(G) ≤ 2n − 2 and that for every pair k, n of integers with
n ≤ k ≤ 2n − 2, there exists a connected graph of order n and diameter two with rn(G) = k.
Also, in the same paper a characterization of connected graphs of order n and diameter two
with prescribed radio number is presented.
In 2002, Ping Zhang [15] discussed upper and lower bounds for a radio number of cycles.
The bounds are showed to be tight for certain cycles. In 2004, Liu and Xie [5] investigated
the radio number of square of cycles. In 2007, B. Sooryanarayana and Raghunath P [12] have
determined radio labeling of cube of a cycle, for all n ≤ 20, all even n ≥ 20 and gave bounds
for other cycles. In [13], they also determined radio number of the graph Cn4 , for all even n and
odd n ≤ 25.
A radio labeling is called radio graceful if rn(G) = n. In [12] and [13] it is shown that the
graph Cn3 is radio graceful if and only if n ∈ {3, 4, 5, 6, 7, 9, 10, 11, 12, 13, 18, 19} and Cn4 is radio
graceful if and only if n ∈ {3, 4, 5, 6, 7, 8, 9, 11, 12, 13, 14, 15, 16, 17, 23, 24, 25}.
In 2005, D. D. F. Liu and X. Zhu [7] completely determined radio numbers of paths and
cycles. In 2006, D. D. F. Liu [8] obtained lower bounds for the radio number of trees, and
characterized the trees achieving this bound. Moreover in the same paper, he gave another
lower bound for the radio number of the trees with at most one vertex of degree more than two
(called spiders) in terms of the lengths of their legs and also characterized the spiders achieving
this bound.
The results of D. D. F. Liu [8] generalizes the radio number for paths obtained by D. D. F.
Liu and X. Zhu in [7]. Further, D.D.F. Liu and M. Xie obtained radio labeling of square of paths
in [6]. In this paper, we completely determine the radio labeling of cube of a path. The main
result we prove in this paper is the following Theorem 1.1. The lower bound is established in
section 2 and a labeling procedure is given in section 3 to show that the lower bounds achieved
in section 3 are really the tight upper bounds.
 Radio Number of Cube of a Path 7

Theorem 1.1 Let Pn3 be a cube of a path on n (n ≥ 6 and n 6= 7) vertices. Then


 n2 +12

 6 , if n ≡ 0 (mod 6)




2
n −2n+19
, if n ≡ 1 (mod 6)

 6


 n2 +2n+10
, if n ≡ 2 (mod 6)
6
rn(Pn3 )= .
 n2 +15

 6 , if n ≡ 3 (mod 6)



 2
n −2n+16

 6 , if n ≡ 4 (mod 6)


 n2 +2n+13
, if n ≡ 5 (mod 6)
6

We recall the following results for immediate reference.

Theorem 1.2(Daphne Der-Fen Liu, Xuding Zhu [6]) For any integer n ≥ 4,

 2k 2 + 3, if n = 2k + 1;
rn(Pn ) =
 2k 2 − 2k + 2 if n = 2k

Lemma 1.3(Daphne Der-Fen Liu, Melanie Xie [7]) Let Pn2 be a square path on n vertices with
k = ⌊ n2 ⌋. Let {x1 , x2 , . . . , xn } be a permutation of V (Pn2 ) such that for any 1 ≤ i ≤ n − 2,

min{dPn (xi , xi+1 ), dPn (xi+1 , xi+2 ) ≤ k + 1,

and if k is even and the equality in the above holds, then dPn (xi , xi+1 ) and dPn (xi+1 , xi+2 )
have different parities. Let f be a function, f : V (Pn2 ) → {0, 1, 2, . . .} with f (x1 ) = 0, and
f (xi+1 ) − f (xi ) = k + 1 − d(xi , xi+1 ) for all 1 ≤ i ≤ n − 1. Then f is a radio-labeling for Pn2 .

§2. Lower Bound

In this section we establish the lower bound for Theorem 1.1. Throughout, we denote a path
on n vertices by Pn , where V (Pn ) = {v1 , v2 , v3 . . , vn } and E(Pn ) ={vi vi+1 | i = 1, 2, ..., n − 1}.
A path on odd length is called an odd path and that of even length is called an even path.

Observation 2.1 By the definition of Pn3 , for any two vertices u, v ∈ V (Pn3 ), we get
l m  
d (u,v)
dPn3 (u, v) = Pn 3 and diam(Pn3 ) = n−13

Observation 2.2 An odd path P2k+1 on 2k + 1 vertices has exactly one center namely vk+1 ,
while an even path P2k on 2k vertices has two centers vk and vk+1 .

For each vertex u ∈ V (Pn3 ), the level of u, denoted by l(u), is the smallest distance in Pn
from u to a center of Pn . Denote the level of the vertices in a set A by L(A).

Observation 2.3 For an even n, the distance between two vertices vi and vj in Pn3 is given by
their corresponding levels as;
8 B. Sooryanarayana, Vishu Kumar M. and Manjula K.

 l m
 |l(vi )−l(vj )| n n+2

 3 , whenever 1 ≤ i, j ≤ 2 or 2 ≤ i, j ≤ n

d(vi , vj ) = (1)

 l m

 l(vi )+l(vj )+1 , otherwise
3

Observation 2.4 For an odd n, the distance between two vertices vi and vj in Pn3 is given by
their corresponding levels as;

 l m
 |l(vi )−l(vj |) n+1 n+1

 3 , whenever 1 ≤ i, j ≤ 2 or 2 ≤ i, j ≤ n

d(vi , vj ) = (2)

 l m

 l(vi )+l(vj ) ,
3 otherwise

Observation 2.5 If n is even, then


L(V (Pn3 )) = n2 − 1, n2 − 2, . . . , 2, 1, 0, 0, 1, 2, . . . , n2 − 2, n2 − 1 .

Therefore h i n nn o n2 − 2n
X n
l(vi ) = 2 1 + 2 + · · · + − 1 = −1 = (3)
2 2 2 4
vi ∈V (Pn3 )

Observation 2.6 If n is odd, then


L(V (Pn3 ) = n−1 n−1
2 , 2 − 1, . . . 2, 1, 0, 1, 2, . . .
n−1
2 − 1, n−1
2 .

Therefore
X    
n−1 n−1 n+1 n2 − 1
l(vi ) = 2 1 + 2 + · · · + = = (4)
2 2 2 4
vi ∈V (Pn3 )

Let f be a radio labeling of the graph Pn3 . Let x1 , x2 , . . . , xn be the sequence of the
vertices of Pn3 such that f (xi+1 ) > f (xi ) for every i, 1 ≤ i ≤ n − 1. Then we have

f (xi+1 ) − f (xi ) ≥ diam(Pn3 ) + 1 − d(xi+1 , xi ) (5)

for every i, 1 ≤ i ≤ n − 1.
Summing up n − 1 inequalities in (5), we get
n−1
X n−1
X n−1
X
[f (xi+1 ) − f (xi )] ≥ [diam(Pn3 ) + 1] − d(xi+1 , xi ) (6)
i=1 i=1 i=1

The terms in the left hand side of the inequality (6) cancels each other except the first and
the last term, therefore, inequality (6) simplifies to
n−1
X
f (xn ) − f (x1 ) ≥ (n − 1)[diam(Pn3 ) + 1] − d(xi+1 , xi ) (7)
i=1
 Radio Number of Cube of a Path 9

If f is a minimal radio labeling of Pn3 , then f (x1 ) = 1 (else we can reduce the span of f
by f (xn ) − f (x1 ) + 1 by reducing each label by f (x1 ) − 1). Therefore, inequality (7) can be
written as
n−1
X
f (xn ) ≥ (n − 1)[diam(Pn3 ) + 1] − d(xi+1 , xi ) + 1 (8)
i=1
By the observations 2.3 and 2.4, for every i, 1 ≤ i ≤ n − 1 it follows that

 
l(xi+1 ) + l(xi ) + 1 l(xi+1 ) + l(xi )
d(xi+1 , xi ) ≤ ≤ +1 (9)
3 3
whenever n is even. And

 
l(xi+1 ) + l(xi ) l(xi+1 ) + l(xi ) 2
d(xi+1 , xi ) ≤ ≤ + (10)
3 3 3
whenever n is odd.
Inequalities (9) and (10), together gives,
n−1
X X
n−1
l(xi+1 ) + l(xi )

d(xi+1 , xi ) ≤ +k
i=1 i=1
3
2
where k = 1, if n is even and k = 3 if n is odd.

n−1
X Xn
1 1
⇒ d(xi+1 , xi ) ≤ ×2 l(xi ) − [l(xn ) + l(x1 )] + k(n − 1)
i=1
3 i=1
3

n−1
X n
2X 1
⇒ d(xi+1 , xi ) ≤ l(xi ) + k(n − 1) − [l(x1 ) + l(xn )] (11)
i=1
3 i=1 3

From the inequalities 8 and 11, we get

n
2X 1
f (xn ) ≥ (n − 1)[diam(Pn3 ) + 1] − l(xi ) + 1 − k(n − 1) + [l(x1 ) + l(xn )]
3 i=1 3
n
2X 1
⇒ f (xn ) ≥ (n − 1)diam(Pn3 ) − l(xi ) + 1 + (1 − k)(n − 1) + [l(x1 ) + l(xn )] (12)
3 i=1 3
We now observe that the equality between the second and third terms in (9) holds only if
l(xi+1 ) + l(xi ) ≡ 0 (mod 3) and the equality between the second and third terms in holds only
if(10) l(xi+1 ) + l(xi ) ≡ 1 (mod 3). Therefore, there are certain number of pairs (xi+1 , xi ) for
which the strict inequality holds. That is, the right hand side of (9) as well as (10) will exceed
by certain amount say ξ. Thus, the right hand side of (12) can be refined by adding an amount
ξ as; & '
Xn
2
f (xn ) ≥ (n − 1)diam(Pn3 ) − l(xi ) + 1 + (1 − k)(n − 1) + η + ξ (13)
3 i=1
10 B. Sooryanarayana, Vishu Kumar M. and Manjula K.

1
where η = 3 [l(xi ) + l(xi+1 )].

We also see that the value of ξ increases heavily if we take a pair of vertices on same side of
the central vertex. So, here onwards we consider only those pairs of vertices on different sides
of a central vertex.

Observation 2.7 All the terms in the right side of the inequality (13), except ξ and η, are
the constants for a given path Pn . Therefore, for a tight lower bound these quantities must be
minimized. If n is even, we have two cental vertices and hence a minimal radio labeling will start
the label from one of the central vertices and end at the other vertex, so that l(x1 ) = l(xn ) = 0.
However, if n is odd, as the graph Pn has only one central vertex, either l(x1 ) > 0 or l(xn ) > 0.
Thus, η ≥ 0 for all even n, and η ≥ 13 for all odd n.

The terms η and ξ included in the inequality (13) are not independent. The choice of
initial and final vertices for a radio labeling decides the value of η, but at the same time it (this
choice) also effect ξ (since ξ depends on the levels in the chosen sequence of vertices). Thus,
for a minimum span of a radio labeling, the sum η + ξ to be minimized rather than η or ξ.

Observation 2.8 For each j, 0 ≤ j ≤ 2, define Lj = {v ∈ V (Pn3 )|l(v) ≡ j(mod 3)} and for
each pair (xi+1 , xi ), 1 ≤ i ≤ n − 1 of vertices of V (Pn3 ), let
n o l m
l(xi+1 )+l(xi )
ξi = 3 + 1 − l(xi+1 )+l(x
3
i )+1
, if n is even, or
n o l m
l(xi+1 )+l(xi )
ξi = 3 + 32 − l(xi+1 3)+l(xi ) , if n is odd.

Then there are following three possible cases:

Possibility 1: Either (i) both xi+1 , xi ∈ L0 or (ii) one of them is in L1 and the
other is in L2 . In this case

 0, if n is even
ξi = .
 2 , if n is odd
3

Possibility 2: Either (i) both xi+1 , xi ∈ L2 or (ii) one of them is in L0 and the
other is in L1 . In this case

 1
3, if n is even
ξi = .
 0, if n is odd

Possibility 3: Either (i) both xi+1 , xi ∈ L1 or (ii) one of them is in L0 and the
other is in L2 . In this case

 2
3, if n is even
ξi = .
 1
3, if n is odd
 Radio Number of Cube of a Path 11

Observation 2.9 For the case n is even, the Possibility 1 given in the Observation 2.8 holds for
every pair of consecutive vertices in the sequence of the form either lα1 , rα2 , lα3 , rα4 , lα5 , rα6 , . . . ,
or lβ1 , rγ1 , lβ2 , rγ2 , lβ3 , rγ3 , . . . ,, where lαi , lβi , lγi denote the vertices in the left of a central vertex
and at a level congruent to 0, 1, 2 under modulo 3 respectively, and, rαi ,rβi , rγi denote the
corresponding vertices in the right side of a central vertex of the path Pn . The first sequence
covers only those vertices of Pn3 which are at a level congruent to 0 under modulo 3, and,
the second sequence covers only those vertices of L1 (or L2 ) which lie entirely on one side of
a central vertex. Now, as the sequence x1 , x2 , . . . , xn covers the entire vertex set of Pn3 , the
sequence should have at least one pair as in Possibility 2 (taken this case for minimum ξi ) to
link a vertex in level congruent to 0 under modulo 3 with a vertex not at a level congruent to
0 under modulo 3. For this pair ξi ≥ 13 . Further, to cover all the left as well as right vertices in
the same level congruent to i, 1 ≤ i ≤ 2, we again require at least one pairs as in Possibility 2
or 3. Thus, for this pair again we have ξi ≥ 31 . Therefore,
n
X 2
ξ= ξi ≥
i=1
3
for all even n.

The above Observation 2.9 can be visualize in the graph called level diagram shown in
Figures 1 and 2. A Hamilton path shown in the diagram indicates a sequence x1 , x2 , . . . , xn
where thin edges join the pair of vertices as in Possibility 1 indicted in Observation 2.8 and
the bold edges are that of Possibility 2 or 3. Each of the subgraphs G0,0 , G1,2 and G2,1 is a
complete bipartite graph having only thin edges and s = ⌈ n−46 ⌉.

lα1 rα1

lα1 lα 2 rα 2
rα1
lα 2 rα 2 lα 3 rα 3
lα 3 rα 3

lα s rα s
lα s rα s

G0,0
G0,0

lβ1 rγ 1 lγ 1 rβ1
lβ1 rγ 1 lγ 1 rβ1
lβ 2 rγ 2 lγ 2 rβ 2
lβ 2 rγ 2 lγ 2 rβ 2
lβ 3 rγ 3 lγ 3 rβ 3
lβ 3 rγ 3 lγ 3 rβ 3

lβ s rγ s lγ s rβ s lβ s rγ s lγ s rβ s

G1,2 G2,1 G1,2 G2,1

Figure 1: For Pn3 when n ≡ 0 or 2 (mod 6). Figure 2: For Pn3 when n ≡ 0 or 2 (mod 6).

If ξ = 23 and n ≡ 0 (mod 6), then only two types of Hamilton paths are possible as shown
in Figures 1 and 2. In each of the case either l(x1 ) > 0 or l(xn ) > 0, therefore η ≥ 13 . Hence
η + ξ ≥ 1 in this case.
If η = 0, then both the starting and the ending vertices should be in the subgraph G0,0 .
Thus, one of the thin edges in G0,0 to be broken and one of its ends to be joined to a vertex in
G1,2 and the other to a vertex in G2,1 with bold edges. These two edges alone will not connect
the subgraphs, so to connect G1,2 and G2,1 we need at least one more bold edge. Therefore,
ξ ≥ 1 and hence η + ξ ≥ 1 in this case also.
 12 B. Sooryanarayana, Vishu Kumar M. and Manjula K.

In all the other possibilities for the case n ≡ 0 or 2 under modulo 6, we have η ≥ 31 and
2
ξ≥ so clearly η + ξ ≥ 1.
3,
The situation is slightly different for the case when n ≡ 4 (mod 6). In this case;
If ξ = 23 , then there is one and only one possible type of Hamilton path as shown in Figure
3, so l(x1 ) > 0 and l(xn ) > 0 implies that η ≥ 32 and hence η + ξ ≥ 34 .
Else if, η = 0, then two bold edges are required. One edge is between a vertex of G1,2 and
a vertex of G0,0 , and, the other edge between a vertex of G2,1 and a vertex of G0,0 (for each
such edges ξi ≥ 13 ). These two edges will not connect all the subgraphs. For this, we require
an edge between a vertex of G1,2 and a vertex of G2,1 , which can be done minimally only by
an edge between a pair of vertices as in Possibility 3 indicated in observation 2.8 (for such an
edge ξi = 32 ). Thus, ξ ≥ 2 × 13 + 32 = 34 .
If ξ = 1, then the possible Hamilton path should contain at least either (i) one edge between
G1,2 and G2,1 , and, another edge from G0,0 , or, (ii) three edges from G0,0 . The first case is
impossible because we can not join the vertices that lie on the same side of a central vertex
with ξ = 1 and the second case is possible only if η ≥ 13 .
Hence, for all even n, we get

η + ξ ≥ 1, if n ≡ 0 or 2 (mod 6) (14)
4
η + ξ ≥ , if n ≡ 4 (mod 6) (15)
3

lα1 rα1
lα 2 rα 2 C0
lα 3 rα 3

l β1 rβ1
lα s rα s lα1
lβ 2 rα1
rβ 2
rα 2 lα 2
G0,0

rα k lα k
lβ1 lγ 1
lβ s rβ s
rγ 1 rβ1
lβ 2 rγ 2 lγ 2 rβ 2 G1,0 G0,1
lβ 3 rγ 3 lγ 3 rβ 3
lγ 1 rγ 1
lγ 2 rγ 2
lβ s −1 rγ s −1 lγ s −1 rβ s −1
lβ s rβ s
G1,2 G2,1
lγ s rγ s

G2,2

Figure 3: A Hamilton path in a level

graph for the case n ≡ 4 (mod 6) Figure 4: Level graph for the case n ≡ 3 or 5 (mod 6).

Observation 2.10 For the case n is odd, the Possibility 2 given in observation 2.8 holds for
every pair of consecutive vertices in the sequence of the form lβ1 , rα1 , lβ2 , rα2 , lβ3 , rα3 , . . . , or
rβ1 , lα1 , rβ2 , lα2 , rβ3 , lα3 , . . . , or lγ1 , rγ1 and lγ2 , rγ2 , lγ3 , rγ3 , . . . , where lαi , lβi , lγi denote
the vertices in the left of a central vertex and at a level congruent to 0, 1, 2 under modulo 3
respectively, and, rαi ,rβi and rγi denote the corresponding vertices in the right side of a central
vertex of the path Pn . Let C0 be the central vertex. Then C0 can be joined to one of the first
two sequences or the first sequence can be combined with second sequence through C0 . The
third sequence covers only those vertices of Pn3 which are at a level congruent to 2 under modulo
 Radio Number of Cube of a Path 13

3, and, the first two sequences are disjoint. Hence to get a Hamilton path x1 , x2 , . . . , xn to
cover the entire vertex set of Pn3 , it should have at least a pair as in Possibility 3, (if the vertex
C0 combines first and second sequences) or at least two pairs that are not as in Possibility 1.
Therefore, as the graph contains only one center vertex,
1 1
η≥ and ξ ≥
3 3
The above observation 2.10 will be visualized in the Figure 4.
In either of the cases, we claim that η + ξ ≥ 35
We note here that, if we take more than three edges amongst G1,0 , G0,1 and G2,2 in the
level graphs shown in Figure 4, then ξ ≥ 4 × 13 , so the claim follows immediately as η ≥ 31 .

Case 1: If η = 13 , then l(x1 ) = 0, so the vertex C0 is in either first sequence or in the second
sequence (as mentioned in the Observation 2.10), but not in both. Hence at least two edges are
required to get a Hamilton path. The minimum possible edges amongst G1,0 , G0,1 and G2,2 )
are discussed in the following cases.

Subcase 1.1: With two edges

The only possible two edges (in the sense of minimum ξ) are shown in Figure 5. Thus,
2
ξ ≥ 3 + 32 = 34 . Hence the claim.

C0 C0

l β1 rβ1 l β1 rβ1
lβ 2 rα1 lα1 lβ 2 rα1 lα1
rβ 2 rβ 2
rα 2 lα 2 rα 2 lα 2

rα k lα k rα k lα k
lβ s rβ s lβ s rβ s
G1,0 G0,1 G1,0 G0,1

lγ 1 rγ 1 lγ 1 rγ 1
lγ 2 rγ 2 lγ 2 rγ 2

lγ s rγ s lγ s rγ s

G2,2 G2,2

Figure 5: Level graph(n ≡ 3 or 5 (mod 6)). Figure 6: hamilton cycle(η = 13 , n ≡ 3, 5(mod 6)).

Subcase 1.2: With three edges

The only possible three edges are shown in Figures 5 and 6. In each case, ξ ≥ 43 . Hence
the claim.

Case 2: If η = 32 , then either l(x1 ) = 0 and l(xn ) = 2, or, l(x1 ) = 1 and l(xn ) = 1. In
the first case at least two edges are necessary, both these edges can not be as in Possibility 3
(because two such edges disconnect G0,1 or disconnect G2,2 or form a tree with at least three
end vertices as shown in Figure 7. Similar fact holds true for the second case also (Follows
easily from Figure 8.
14 B. Sooryanarayana, Vishu Kumar M. and Manjula K.

C0 C0

l β1 rβ1 l β1 rβ1
lβ 2 rα1 lα1 lβ 2 rα1 lα1
rβ 2 rβ 2
rα 2 lα 2 rα 2 lα 2

rα k lα k rα k lα k
lβ s rβ s lβ s rβ s
G1,0 G0,1 G1,0 G0,1

lγ 1 rγ 1 lγ 1 rγ 1
lγ 2 rγ 2 lγ 2 rγ 2

lγ s rγ s lγ s rγ s

G2,2 G2,2

1 2
Figure 7: hamilton cycle for the case η = 3 Figure 8: hamilton cycle for the case η = 3 and
and n ≡ 3 or 5(mod 6). n ≡ 3 or 5(mod 6).

1
Hence ξ ≥ 3 + 32 . Therefore,

5
ξ+η ≥ f or n ≡ 3 or 5 (mod 3) (16)
3
The case n ≡ 1(mod 3) follows similarly.

We now prove the necessary part of the Theorem 1.1.

Case 1: n ≡ 0 (mod 6) and n ≥ 6

Substituting the minimum possible bound for η + ξ = 1 (as in equation (14)) diam(Pn3 ) =
Pn n2 −2n
⌈ n−1 n
3 ⌉ = 3, i=1 l(xi ) = 4 (follows by Observation 2.5) and k = 1 in the inequality (13),
we get
   
n 2 n2 − 2n
f (xn ) ≥ (n − 1) − +1+1
3 3 4
 
n2 n2
⇒ f (xn ) ≥ +2 = +2 (17)
6 6
n2 +12
Hence rn(Pn3 ) ≥ 6 , whenever n ≡ 0 (mod 6) and n ≥ 6.

Case 2: n ≡ 1 (mod 6) and n ≥ 13

Substituting the minimum possible bound for η + ξ = 53 (as in equation (14)), diam(Pn3 ) =
Pn n2 −1
⌈ n−1 n−1
3 ⌉= 3 , i=1 l(xi ) = 4 and k = 23 in the inequality (13), we get
     
n−1 2 n2 − 1 1 5
f (xn ) ≥ (n − 1) − + 1 + (n − 1) +
3 3 4 3 3
   
n−1 2 n2 − 1 1 n2 − 2n + 19
⇒ f (xn ) ≥ (n − 1) − + (n − 1) + 3 = (18)
3 3 4 3 6
 Radio Number of Cube of a Path 15

n2 −2n+19
Hence rn(Pn3 ) ≥ 6 , whenever n ≡ 1 (mod 6) and n ≥ 13.

Case 3: n ≡ 2 (mod 6) and n ≥ 8

Substituting the minimum possible bound for η + ξ = 1 (as in equation (14)), diam(Pn3 ) =
Pn n2 −2n
⌈ n−1
3 ⌉= n+1
3 , i=1 l(xi ) = 4 and k = 1 in the inequality (13), we get
   
n+1 2 n2 − 2n
f (xn ) ≥ (n − 1) − +1+1
3 3 4
 
(n − 2)2 (n − 2)2 n2 + 2n + 10
⇒ f (xn ) ≥ +n+1 = +n+1= (19)
6 6 6
n2 +2n+10
Hence rn(Pn3 ) ≥ 6 , whenever n ≡ 2 (mod 6) and n ≥ 8.

Case 4: n ≡ 3 (mod 6) and n ≥ 9

n Pn n2 −1
Substituting η + ξ = 53 , diam(Pn3 ) = ⌈ n−1
3 ⌉ = 3, i=1 l(xi ) = 4 and k = 2
3 in the
inequality (13), we get
   
n 2 n2 − 1 1 5
f (xn ) ≥ (n − 1) − + 1 + (n − 1) +
3 3 4 3 3
 
(n − 3)2 (n − 3)2 n2 + 15
⇒ f (xn ) ≥ +n+1 = +n+1= (20)
6 6 6
n2 +15
Hence rn(Pn3 ) ≥ 6 , whenever n ≡ 3 (mod 6) and n ≥ 9.

Case 5: n ≡ 4 (mod 6) and n ≥ 10

Substituting the minimum possible bound for η + ξ = 43 (as in equation (15)), diam(Pn3 ) =
Pn n2 −2n
⌈ n−1
3 ⌉= n−1
3 , i=1 l(xi ) = 4 and k = 1 in the inequality (13), we get
   
n − 1 2 n2 − 2n 4
f (xn ) ≥ (n − 1) − +1+
3 3 4 3
 
(n − 4)2 (n − 4)2 n2 − 2n + 16
f (xn ) ≥ +n = +n= (21)
6 6 6
n2 −2n+16
Hence rn(Pn3 ) ≥ 6 , whenever n ≡ 4 (mod 6) and n ≥ 10.

Case 6: n ≡ 5 (mod 6) and n ≥ 11

Pn n2 −1
Substituting η + ξ = 53 , diam(G) = ⌈ n−1
3 ⌉ =
n+1
3 , i=1 l(xi ) = 4 and k = 2
3 in the
inequality (13), we get
     2 
n + 1 2 n2 − 1 1 5 n + 2n + 13
f (xn ) ≥ (n − 1) − + 1 + (n − 1) + =
3 3 4 3 3 6
 
(n − 5)2 (n − 5)2 n2 + 2n + 13
⇒ f (xn ) ≥ + 2(n − 1) = + 2(n − 1) = (22)
6 6 6
n2 +n+13
Hence rn(Pn3 ) ≥ 6 , whenever n ≡ 5 (mod 6) and n ≥ 11.
16 B. Sooryanarayana, Vishu Kumar M. and Manjula K.

§3. Upper Bound and Optimal Radio-Labelings

We now establish Theorem 1.1, it suffices to give radio-labelings that achieves the desired spans.
Further, we will prove the following lemma similar to the Lemma 1.3 of Daphne Der-Fen Liu
and Melanie Xie obtained in [7].

 n−1 
Lemma 3.1 Let Pn3 be a cube path on n (n ≥ 6) vertices with k = 3 . Let {x1 , x2 , . . . , xn }
be a permutation of V (Pn3 ) such that for any 1 ≤ i ≤ n − 2,

min{dPn (xi , xi+1 ), dPn (xi+1 , xi+2 )} ≤ 3 k2 + 1

if k is even and the equality in the above holds, then the sum of the parity congruent to 0 under
modulo 3. Let f be a function, f : V (Pn3 ) → {1, 2, 3, . . .} with f (x1 ) = 1, and f (xi+1 ) − f (xi ) =
k + 1 − d(xi , xi+1 ) for all 1 ≤ i ≤ n − 1, where d(xi , xi+1 ) = dPn3 (xi , xi+1 ). Then f is a
radio-labeling for Pn3 .

Proof Recall, diam(Pn3 ) = k. Let f be a function satisfying the assumption. It suffices to

prove that f (xj ) − f (xi ) ≥ k + 1 − d(xi , xj ) for any j ≥ i + 2. For i = 1, 2, . . . , n − 1, set

fi = f (xi+1 ) − f (xi ).

Since the difference in two consecutive labeling is at least one it follows that fi ≥ 1. Further,
for any j ≥ i + 2, it follows that

f (xj ) − f (xi ) = fi + fi+1 + · · · + fj−1 .

Suppose j = i + 2. Assume d(xi , xi+1 ) ≥ d(xi+1 , xi+2 ). (The proof for d(xi+1 , xi+2 ) ≥
d(xi , xi+1 ) is similar.) Then, d(xi+1 , xi+2 ) ≤ k+2
2 . Let xi = va , xi+1 = vb , and xi+2 = vc .
It suffices to consider the following cases.

Case 1: b < a < c or c < a < b

k+2
Since d(xi , xi+1 ) ≥ d(xi+1 , xi+2 ), we obtain d(xi , xi+1 ) = d(xi+1 , xi+2 ) ≤ 2 and dPn (xi , xi+2 ) ≤
2 so, d(xi , xi+2 ) = 1. Hence,

f (xi+2 ) − f (xi ) = fi + fi+1

= k + 1 − d(xi , xi+1 ) + k + 1 − d(xi+1 , xi+2 )
 
k+2
≥ 2k + 2 − 2
2
= k+1−1
= k + 1 − d(xi , xi+2 )

Case 2: a < b < c or c < b < a

 Radio Number of Cube of a Path 17

In this case, d(xi , xi+2 ) ≥ d(xi , xi+1 ) + d(xi+1 , xi+2 ) − 1 and hence

f (xi+2 ) − f (xi ) = fi + fi+1

= k + 1 − d(xi , xi+1 ) + k + 1 − d(xi+1 , xi+2 )
= 2k + 2 − {d(xi , xi+1 ) + d(xi+1 , xi+2 )}
= 2k + 2 − {d(xi , xi+2 ) + 1}
= 2k + 1 − d(xi , xi+2 )
≥ k + 1 − d(xi , xi+2 )

Case 3: a < c < b or b < c < a

Assume min{dPn (xi , xi+1 ), dPn (xi+1 , xi+2 )} < 3 k2 + 1, then we have d(xi+1 , xi+2 ) < k+2
2
and by triangular inequality,

d(xi , xi+2 ) ≥ d(xi , xi+1 ) − d(xi+1 , xi+2 )

Hence,

f (xi+2 ) − f (xi ) = fi + fi+1

= k + 1 − d(xi , xi+1 ) + k + 1 − d(xi+1 , xi+2 )
= 2k + 2 − [d(xi , xi+1 ) − d(xi+1 , xi+2 )] − 2d(xi+1 , xi+2 )
≥ 2k + 2 − [d(xi , xi+2 )] − 2d(xi+1 , xi+2 )
 
k+2
> 2k + 2 − d(xi , xi+2 ) − 2
2
= k − d(xi , xi+2 )

Therefore,

f (xi+2 ) − f (xi ) ≥ k + 1 − d(xi , xi+2 )

If min{dPn (xi , xi+1 ), dPn (xi+1 , xi+2 )} = 3 k2 + 1, then by our assumption, it must be that
dPn (xi+1 , xi+2 ) = 3 k2 + 1 (so k is even) ,and, sum of dPn (xi , xi+1 ) and dPn (xi+1 , xi+2 ) is
congruent to 0 under modulo 3 implies that dPn (xi , xi+1 ) 6≡ 0 (mod 3). Hence, we have

d(xi , xi+2 ) = d(xi , xi+1 ) − d(xi+1 , xi+2 ) + 1.

This implies

f (xi+2 ) − f (xi ) = fi + fi+1

= 2(k + 1) − [d(xi , xi+2 )] − d(xi+1 , xi+2 ) − d(xi+1 , xi+2 ) + 1
≥ 2k + 2 − 2[d(xi+1 , xi+2 )] − d(xi , xi+2 ) + 1
 
k+2
≥ 2k + 2 − 2 − d(xi , xi+2 ) + 1
2
= k + 1 − d(xi , xi+2 )

Let j = i + 3. First, we assume that the sum of some pairs of the distances d(xi , xi+1 ),
d(xi+1 , xi+2 ), d(xi+2 , xi+3 ) is at most k + 2. Then
18 B. Sooryanarayana, Vishu Kumar M. and Manjula K.

d(xi , xi+1 ) + d(xi+1 , xi+2 ) + d(xi+2 , xi+3 ) ≤ (k + 2) + k = 2k + 2

and hence,

f (xi+3 ) − f (xi ) = 3k + 3 − d(xi , xi+1 ) − d(xi+1 , xi+2 ) − d(xi+2 , xi+3 )

≥ 3k + 3 − (2k + 2)
= k + 1 > k + 1 − d(xi , xi+3 ).

Next, we assume that the sum of every pair of the distances d(xi , xi+1 ), d(xi+1 , xi+2 ) and
d(xi+2 , xi+3 ) is greater than k + 2. Then, by our hypotheses, it follows that

k+2 k+2
d(xi , xi+1 ), d(xi+2 , xi+3 ) ≥ and d(xi+1 , xi+2 ) ≤ (23)
2 2
Let xi = va , xi+1 = vb , xi+2 = vc , xi+3 = vd . Since diam(Pn3 ) = k, by equation
(23) and our assumption that the sum of any pair of the distances, d(xi , xi+1 ), d(xi+1 , xi+2 ),
d(xi+2 , xi+3 ), is greater than k + 2, it must be that a < c < b < d (or d < b < c < a). Then

d(xi , xi+3 ) ≥ d(xi , xi+1 ) + d(xi+2 , xi+3 ) − d(xi+1 , xi+2 ) − 1.

So,

d(xi , xi+1 ) + d(xi+1 , xi+2 ) + d(xi+2 , xi+3 ) ≤ d(xi , xi+3 ) + d(xi+1 , xi+2 ) + 1
k+2
≤ d(xi , xi+3 ) + +1
2
k
= d(xi , xi+3 ) + + 2
2
By equation 23, we have

f (xi+3 ) − f (xi ) = 3k + 3 − d(xi , xi+1 ) − d(xi+1 , xi+2 ) − d(xi+2 , xi+3 )

k
≥ 3k + 3 − 2 − − d(xi , xi+3 )
2
= k + 1 − d(xi , xi+3 ).

Let j ≥ i + 4. Since min{d(xi , xi+1 ), d(xi+1 , xi+2 )} ≤ k+2

2 , and fi ≥ k + 1 − d(xi , xi+1 )
for any i, we have max{fi , fi+1 } ≥ k2 for any 1 ≤ i ≤ n − 2. Hence,

f (xj ) − f (xi ) ≥ fi + fi+1 + fi+2 + fi+3

   
k k
≥ 1+ + 1+
2 2
> k + 1 > k + 1 − d(xi , xj )


To show the existence of a radio-labeling achieving the desired bound, we consider cases
separately. For each radio-labeling f given in the following, we shall first define a permutation
(line-up) of the vertices V (Pn3 ) = {x1 , x2 , . . . , xn }, then define f by f (x1 ) = 1 and for i =
1, 2, . . . , n − 1:
f (xi+1 ) = f (xi ) + diam(Pn3 ) + 1 − dPn3 (xi , xi+1 ). (24)
 Radio Number of Cube of a Path 19

For the case n ≡ 0 (mod 6)

 
Let n = 6p. Then k = n−1 3 = 2p ⇒ k2 = p. Arrange the vertices of the graph Pn3 as
x1 = v3p+1 , x2 = v2 , x3 = v3p+3 , . . . , xn = v3p as shown in the Table 1.
Define a function f by f (x1 ) = 1 and for all i, 1 ≤ i ≤ n − 1,

f (xi+1 ) = f (xi ) + 2p + 1 − dPn3 (xi , xi+1 ) (25)

The function f defined in equation (25) chooses the vertices one from the right of the central
vertex and other from the left for the consecutive labeling. The difference between two adjacent
vertices in Pn is shown above the arrow. Since the minimum of any two consecutive distances
is lesser than 3p + 1 and equal to 3p + 1 only if their sum is divisible by 3, by Lemma 3.1, it
follows that f is a radio labeling.

[3 p −1] [3 p +1 3 p −2 3 p +1 3 p −2
x1 = v3 p +1  
→ v2  → v3 p +3 → v5 → v3 p + 6 → v8

3 p +1 3 p +1 3 p − 2] [3 p +1] [6 p −1]
→ • • • → v6 p −3  
→ v3 p −1  
→ v6 p  
→ v1

[3 p +1 3 p −2 3 p +1 3 p −2 3 p +1 3 p +1
 → v3 p + 2 → v4 → v3 p +5 → v7 → • • • →

3 p − 2] [3 p +1] [ 6 p − 4] [3 p +1 3 p −2
v6 p − 4  
→ v3 p − 2  
→ v6 p −1  
→ v3  → v3 p + 4 →

3 p +1 3 p −2 3 p +1 3 p +1 3 p − 2]
v6 → v3 p + 7 → v9 → • • • → v6 p − 2  
→ v3 p = xn

Table 1: A radio-labeling procedure for the graph Pn3 when n ≡ 0 (mod 6)

For the labeling f defined above we get

Xn        
3p − 1 3p + 1 3p − 2 3p + 1
d(xi , xi+1 ) = + + (p − 1) + +
i=1
3 3 3 3
       
6p − 1 3p + 1 3p − 2 3p + 1
+ + (p − 1) + +
3 3 3 3
     
6p − 4 3p + 1 3p − 2
+ + (p − 1)
3 3 3
= p + (2p + 1)(p − 1) + (p + 1) + 2p + (2p + 1) + (p − 1) +
(p + 1) + (2p − 1) + (2p + 1)(p − 1)
= 6p2 + 4p − 2.

Therefore,
n
X
f (xn ) = (n − 1)(diamP23 + 1) − d(xi , xi+1 ) + f (x1 )
i=1
= (6p − 1)(2p + 1) − (6p2 + 4p − 2) + 1
n2
= 6p2 + 2 = + 2.
2
Hence,
20 B. Sooryanarayana, Vishu Kumar M. and Manjula K.

n2 +12
rn(Pn3 ) ≤ 6 , if n ≡ 0 (mod 6)

An example for this case is shown in Figure 9.

v1 v2 v3 v4 v5 v6

23 5 42 30 12 49
x8 x2 x14 x10 x4 x16
v12 v11 v10 v9 v8 v7

8 26 1 56 19 37
x3 x9 x1 x18 x6 x12
v13 v14 v15 v16 v17 v18

45 33 15 52 40 22
x15 x11 x5 x17 x13 x7

3
Figure 9: A minimal radio labeling of the graph P18 .

For the case n ≡ 1 (mod 6)

 
Let n = 6p + 1. Then k = n−1 3 = 2p ⇒ k2 = p. Arrange the vertices of the graph Pn3 as
x1 = v3p+1 , x2 = v3 , x3 = v3p+4 , . . . , xn = v6p−1 as shown in the Table 2.

Define a function f by f (x1 ) = 1 and for all i, 1 ≤ i ≤ n − 1,

f (xi+1 ) = f (xi ) + 2p + 1 − dPn3 (xi , xi+1 ). (26)

For the function f defined in equation (26), The minimum difference between any two adjacent
vertices in Pn is shown in Table 2 is less than 3p + 1 and equal to 3p + 1 only if their sum is
divisible by 3, by Lemma 3.1, it follows that f is a radio labeling.

[3 p −2 3 p +1 3 p −2 3 p +1 3 p −2
x1 = v3 p +1  
→ v3 → v3 p + 4 → v6 → v3 p + 7 → v9

3 p +1 3 p−2 3 p +1] 6 p −1 [3 p +1 3 p −2
→ • • •  → v3 p  
→ v6 p +1 → v2  
→ v3 p + 3  →

3 p +1 3 p−2 3 p−2 3 p +1] [6 p −1] [3 p +1

v5 → v3 p + 6  → • • •  → v3 p −1  
→ v6 p  
→ v1  
→

3 p −2 3 p +1 3 p −2 3 p − 2] 3 p +1
v3 p + 2  → v4 → v3 p + 5 → • • •  
→ v3 p − 2 → v6 p −1 = xn

Table 2: A radio-labeling procedure for the graph Pn3 when n ≡ 1 (mod 6)

 Radio Number of Cube of a Path 21

For the labeling f defined above we get

n
X          
3p − 3 3p + 1 3p − 2 3p + 1 6p − 4
d(xi , xi+1 ) = + + + (p − 1) + +
i=1
3 3 3 3 3
       
3p + 1 3p − 2 3p + 1 6p − 1
+ (p − 1) + + +
3 3 3 3
         
3p + 1 3p − 2 3p + 1 6p − 1 3p + 1
+ (p − 1) + + +
3 3 3 3 3
= 2p + (2p + 1)(p − 2) + 2p − 1 + (2p + 1)(p − 1) + 3p + 1 + (2p + 1)(p − 1) + 4p + 2
= 6p2 + 6p − 2.

Therefore,
n
X
f (xn ) = (n − 1)(diamP23 + 1) − d(xi , xi+1 ) + f (x1 )
i=1
= (6p)(2p + 1) − (6p2 + 6p − 2) + 1
n2 − 2n + 19
= 6p2 + 3 = .
6
Hence,
n2 −2n+19
rn(Pn3 ) ≤ 6 , if n ≡ 1 (mod 6) and n ≥ 13

An example for this case is shown in Figure 10.

41 23 5 48 30 12 55 37 19
1

22 40 58 15 33 51 8 26 44

3
Figure 10: A minimal radio labeling of the graph P19 .

For the case n ≡ 2 (mod 6)

 
Let n = 6p + 2. Then k = n−1 3 = 2p + 1 ⇒ 3p + 2 < 3 k2 + 1. Arrange the vertices of the
graph Pn3 as x1 = v3p+1 , x2 = v6p+2 , x3 = v3p−2 , . . . , xn = v3p+3 as shown in the Table 3.
Define a function f by f (x1 ) = 1 and for all i, 1 ≤ i ≤ n − 1,

f (xi+1 ) = f (xi ) + 2p + 2 − dPn3 (xi , xi+1 ) (27)

For the function f defined in equation (27), the minimum difference between any two adjacent
vertices in Pn is shown in Table 3 is not greater than 3p + 1 and k is odd, by Lemma 3.1, it
follows that f is a radio labeling.
22 B. Sooryanarayana, Vishu Kumar M. and Manjula K.

[3 p +1 3 p +4 3 p +1 3 p +4 3 p +1
x1 = v3 p +1  
→ v6 p + 2  → v3 p − 2 → v6 p −1  → v3 p − 5 →

3 p +4 3 p +1 3 p + 4] [3 p +1] [ 2]
v6 p − 4  → • • • → v3 p + 5  
→ v1  
→ v3 p + 2 → v3 p

[3 p +1 3 p+4 3 p +1 3 p+4 3 p +1
 → v6 p +1  → v3 p − 3 → v6 p − 2  → v3 p − 6 → v6 p − 5

3 p+4 3 p + 4] [3 p +1] [5] [3 p +1

 → v3 p −9 • • •  
→ v3  
→ v3 p + 4 → v3 p −1  → v6 p

3 p+4 3 p +1 3 p+4 3 p +1 3 p+4

 → v3 p − 4 → v6 p −3  → v3 p − 7 → v6 p − 6 • • •  → v5

3 p +1 3 p + 4] 3 p +1
→ v3 p + 6  
→ v2 → v3 p +3 = xn

Table 3: A radio-labeling procedure for the graph Pn3 when n ≡ 2 (mod 6)

For the labeling f defined above we get

n
X        
3p + 1 3p + 4 3p + 1 2
d(xi , xi+1 ) = + (p) + + +
i=1
3 3 3 3
       
3p + 1 3p + 4 3p + 1 5
+ (p − 1) + + +
3 3 3 3
     
3p + 1 3p + 4 3p + 1
+ (p − 1) +
3 3 3
= (2p + 3)(p) + (p + 1) + 1 + (2p + 3)(p − 1) + (p + 1) +
2 + (2p + 3)(p − 1) + (p + 1)
= 6p2 + 8p.

Therefore,

22 71 47 15 64 40 8 57 33 1

5 37 61 12 44 68 19 51 75 26

3
Figure 11: A minimal radio labeling of the graph P20 .
 Radio Number of Cube of a Path 23

n
X
f (xn ) = (n − 1)(diamP23 + 1) − d(xi , xi+1 ) + f (x1 )
i=1
= (6p + 1)(2p + 2) − (6p2 + 8p) + 1
n2 + 2n + 10
= 6p2 + p + 3 = .
6

Hence,

n2 +2n+10
rn(Pn3 ) ≤ 6 ., if n ≡ 2 (mod 6)

An example for this case is shown in Figure 11.

For the case n ≡ 3 (mod 6)

 
Let n = 6p + 3. Then k = n−1 3 = 2p + 1 ⇒ 3p + 2 < 3 k2 + 1. Arrange the vertices of the
3
graph Pn as x1 = v3p+1 , x2 = v6p+2 , x3 = v1 , . . . , xn = v3p+4 as shown in the Table 4.
Define a function f by f (x1 ) = 1 and for all i, 1 ≤ i ≤ n − 1,

f (xi+1 ) = f (xi ) + 2p + 2 − dPn3 (xi , xi+1 ) (28)

For the function f defined in equation (28), the minimum difference between any two adjacent
vertices in Pn is shown in Table 4 is not greater than 3p + 1 and k is odd, by Lemma 3.1, it
follows that f is a radio labeling.

[3 p +1] [6 p +1] [3 p +1] [3 p +1] [6 p +1]

x1 = v3 p +1  
→ v6 p + 2  
→ v1  
→ v3 p + 2  
→ v6 p +3  
→

[3 p +1 3 p −2 3 p +1 3 p −2 3 p +1
v2  → v3 p + 3 → v5 → v3 p + 6 → v8 → v3 p + 9 • • •

3 p − 2] [3 p +1] [ 6 p − 4] [3 p +1 3 p −2
v6 p −3  
→ v3 p −1  
→ v6 p  
→ v4  → v3 p +5 → v7

3 p +1 3 p −2 3 p − 2] [3 p +1] [3 p −1]
→ v3 p +8 → • • • v6 p − 4  
→ v3 p − 2  
→ v6 p −1  
→

[3 p +1 3 p+4 3 p +1 3 p +4 3 p +1
v3 p  → v6 p +1 → v3 p −3 → v6 p − 2 → v3 p −6 →

3 p + 4] [3 p +1]
v6 p −5 • • • v3 p + 7  
→ v3  
→ v3 p + 4 = xn

Table 4: A radio-labeling procedure for the graph Pn3 when n ≡ 3 (mod 6).
24 B. Sooryanarayana, Vishu Kumar M. and Manjula K.

For the labeling f defined above we get

Xn        
3p + 1 6p + 1 3p + 1 6p + 1
d(xi , xi+1 ) = + + + +
i=1
3 3 3 3
       
3p + 1 3p − 2 3p + 1 6p − 4
+ (p − 1) + + +
3 3 3 3
     
3p + 1 3p − 2 3p + 1
+ (p − 2) + +
3 3 3
       
3p − 1 3p + 1 3p − 4 3p + 1
+ + (p − 1) +
3 3 3 3
= (p + 1) + (2p + 1) + (p + 1) + (p + 1)(2p + 1) +
(2p + 1)(p − 1) + (p + 1) + (2p − 1) +
(2p + 1)(p − 2) + (p + 1) + p + (2p + 3)(p − 1) + (p + 1)
= 6p2 + 10p + 1.

Therefore,
n
X
f (xn ) = (n − 1)(diamP23 + 1) − d(xi , xi+1 ) + f (x1 )
i=1
= (6p + 2)(2p + 2) − (6p2 + 10p + 1) + 1
n2 + 15
= 6p2 + 6p + 4 = .
6
Hence,
n2 +15
rn(Pn3 ) ≤ 6 , if n ≡ 3 (mod 6)

An example for this case is shown in Figure 12.

6 15 72 40 24 65 49 33 58 1

10

14 5 62 37 53 69 28 44 76 19

3
Figure 12: A minimal radio labeling of the graph P21

For the case n ≡ 4 (mod 6)

 
Let n = 6p + 4. Then k = n−1 3 = 2p + 1 ⇒ 3p + 2 ≤ 3 k2 + 1. Arrange the vertices of the
graph Pn3 as x1 = v3p+1 , x2 = v6p+2 , x3 = v3p−2 , . . . , xn = v6p+4 as shown in the Table 5.
Define a function f by f (x1 ) = 1 and for all i, 1 ≤ i ≤ n − 1,

f (xi+1 ) = f (xi ) + 2p + 2 − dPn3 (xi , xi+1 ). (29)

 Radio Number of Cube of a Path 25

For the function f defined in equation (29), the minimum difference between any two adjacent
vertices in Pn is shown in Table 5 is not greater than 3p + 4 and k is odd, by Lemma 3.1, it
follows that f is a radio labeling.

[3 p +1] [3 p + 4] [3 p +1 3 p +4
x1 = v3 p +1  
→ v6 p + 2  
→ v3 p − 2  → v6 p −1 → v3 p −5

3 p +1 3 p +4 3 p +1 3 p +1 3 p + 4]
→ v6 p − 4 → v3 p −8 → • • • → v3 p + 5  
→ v1

[3 p + 2] [3 p +1 3 p+4 3 p +1 3 p +4
 
→ v3 p +3  → v2 → v3 p + 6 → v5 → • • •

3 p+4 3 p +1 3 p +1] [3 p +1] [3 p + 2]

→ v6 p → v3 p −1  → v6 p +3  
→ v3 p + 2  
→ v6 p + 4

[3 p + 4 3 p +1 3 p +4 3 p +1 3 p+4
 
→ v3 p → v6 p +1 → v3 p −3 → v6 p − 2 → v3 p −6

3 p +1 3 p+4 3 p +1]
→ • • • v3 p + 7 → v3  → v3 p + 4 = xn

Table 5: A radio-labeling procedure for the graph Pn3 when n ≡ 4 (mod 6).

For the labeling f defined above we get

n
X        
3p + 1 3p + 4 3p + 1 3p + 4
d(xi , xi+1 ) = + + + (p − 1) +
i=1
3 3 3 3
       
3p + 1 3p + 4 3p + 1 3p + 2
+ (p) + + +
3 3 3 3
   
3p + 4 3p + 1
+ (p)
3 3
= (p + 1) + (p + 2) + (2p + 3)(p − 1) + (p + 1) + (2p + 3)p +
(p + 1) + (p + 1) + (2p + 3)p = 6p2 + 12p + 3.

Therefore,
n
X
f (xn ) = (n − 1)(diamP23 + 1) − d(xi , xi+1 ) + f (x1 )
i=1
= (6p + 3)(2p + 2) − (6p2 + 12p + 3) + 1
n2 − 2n + 16
= 6p2 + 6p + 4 = .
6
Hence,
n2 −2n+16
rn(Pn3 ) ≤ 6 , if n ≡ 4 (mod 6)

An example for this case is shown in Figure 13.

For the case n ≡ 5 (mod 6)

 
Let n = 6p + 5. Then k = n−1 3 = 2p + 2 ⇒ 3p + 4 ≤ 3 k2 + 1. Arrange the vertices of the
graph Pn3 as x1 = v3p+3 , x2 = v2 , x3 = v3p+6 , . . . , xn = v3p+4 as shown in the Table 6.
26 B. Sooryanarayana, Vishu Kumar M. and Manjula K.

11 17 37 6 22 32 1 27

30 24 4 35 19 9 40 14

3
Figure 13: A minimal radio labeling of the graph P16 .

Define a function f by f (x1 ) = 1 and for all i, 1 ≤ i ≤ n − 1,

f (xi+1 ) = f (xi ) + 2p + 3 − dPn3 (xi , xi+1 ) (30)

For the function f defined in equation (30), the maximum difference between any two adjacent
vertices in Pn is shown in Table 6 is less than or equal to 3p + 4 and the equality holds only if
their sum is divisible by 3, by Lemma 3.1, it follows that f is a radio labeling.

[3 p +1 3 p+4 3 p +1 3 p+ 4 3 p +1
x1 = v3 p +3  → v2 → v3 p + 6 → v5 → v3 p +9 → • • •

3 p +1 3 p + 4] [3 p +1] [3 p +3] [3 p + 4
→ v3 p −1  
→ v6 p +3  
→ v3 p + 2  
→ v6 p +5  
→ v3 p +1

3 p +1 3 p+4 3 p +1 3 p+4 3 p+4

→ v6 p + 2 → v3 p − 2 → v6 p −1 → • • • → v4

3 p +1] [3 p + 4] [6 p +3] [3 p + 4 3 p +1
 → v3 p +5  
→ v1  
→ v6 p + 4  
→ v3 p → v6 p +1

3 p+4 3 p +1 3 p +4 3 p +1 3 p +1
→ v3 p −3 → v6 p − 2 → v3 p −6 → • • • → v3 p + 7

3 p+4 3 p +1]
→ v3  → v3 p + 4 = xn

Table 6: A radio-labeling procedure for the graph Pn3 when n ≡ 5 (mod 6)

For the labeling f defined above we get

X n        
3p + 1 3p + 4 3p + 1 3p + 3
d(xi , xi+1 ) = + (p) + + +
i=1
3 3 3 3
       
3p + 4 3p + 1 3p + 4 6p + 3
+ (p) + + +
3 3 3 3
   
3p + 4 3p + 1
+ (p)
3 3
= (2p + 3)p + (p + 1) + (p + 1) + (2p + 1)p + (p + 2) +
(2p + 1) + (2p + 3)(p) = 6p2 + 14p + 5.
 Radio Number of Cube of a Path 27

Therefore,
n
X
f (xn ) = (n − 1)(diamP23 + 1) − d(xi , xi+1 ) + f (x1 )
i=1
= (6p + 4)(2p + 3) − (6p2 + 14p + 5) + 1
n2 + 2n + 13
= 6p2 + 12p + 8 = .
6
Hence,
n2 +2n+13
rn(Pn3 ) ≤ 6 , if n ≡ 5 (mod 6)

An example for this case is shown in Figure 14.

69 6 93 60 15 84 51 24 75 42 33

38 71 28 47 80 19 56 89 10 65 98

3
Figure 14: A minimal radio labeling of the graph P23 .

§4. Radio labeling of Pn3 for n ≤ 5 or n = 7

In this section we determine radio numbers of cube path of small order as a special case.

Theorem 4.1 For any integer n, 1 ≤ n ≤ 5, the radio number of the graph Pn3 is given by

 n, if n = 1, 2, 3, 4
rn(Pn2 ) =
 8, if n = 5, 7

Proof If n ≤ 4, the graph is isomorphic to Kn and hence the result follows immediately.
Now consider the case n = 5, we see that there is exactly one pair of vertices at a distance 2
P
and all other pairs are adjacent, so maximum value of 41 d(xi , xi+1 ) = 2 + 1 + 1 + 1 = 5.
Now, consider a radio labeling f of P53 and label the vertices as x1 , x2 , x3 , x4 , x5 such that
f (xi ) < f (xi+1 ), then
4
X
f (xn ) − f (x1 ) ≥ (n − 1)(diamP53 + 1) − d(xi , xi+1 )
1
≥ 4(3) − 5 = 7
⇒ f (xn ) ≥ 7 + f (x1 ) = 8
28 B. Sooryanarayana, Vishu Kumar M. and Manjula K.

4 8

3 6

Figure 15: A minimal radio labeling of P53 .

includegraphics[width=8cm]figlast2.eps

Figure 16: A minimal radio labeling of P73

On the other hand, In the Figure 15, we verify that the labels assigned for the vertices serve as
a radio labeling with span 8, so rn(Pn3 ) = 8.
, similarly if n = 7, then, as the central vertex of Pn3 is adjacent to every other vertex,
P
maximum value of 6 i = 1d(xi , xi+1 ) = 2 × 5 + 1 = 11. So, as above, f (xn ) ≥ (6)(2 + 1) −
11 + 1 = 8. The reverse inequality follows by the Figure 16. Hence the theorem. 

Acknowledgment

Authors are very much thankful to the referees for their valuable suggestions for the improve-
ment of the paper and the Principals Prof. Martin Jebaraj of Dr. Ambedkar Institute of
Technology Prof. Rana Prathap Reddy of Reva Institute of Technology and Prof. K. R. Suresh
of Bangalore Institute of Technology, for their constant support and encouragement during the
preparation of this paper.

References

[1] Buckley F and Harary F, Distance in Graphs, Addison-Wesley,(1990).

[2] G. Chartrand, D. Erwin, F. Harary and P. Zhang, Radio labelings of graphs, Bull. Inst.
Combin. Appl., 33 (2001), 77-85.
[3] G. Chartrand, D. Erwin, F. Harary and P. Zhang, Radio Antipodal Colorings of Cycles,
Congr. Numer, 144 (2000),129-141.
[4] Gary Chartrand, David Erwin, Ping Zhang, A graph labeling problem Suggested by FM
Channel Restrictions, Bulletin of the Inst. Combin. Appl. 43 (2005), 43-57.
[5] D. Liu and M. Xie, Radio number for square of cycles, congr, Numer., 169 (2004) 105-125.
[6] D.D.F. Liu and M. Xie, Radio number of square paths, Ars Comb., 90(2009), 307-319.
[7] Daphne Der-Fen Liu and Xuding Zhu, Multilevel Distance Labelings for paths and cycles,
SIAM J. Discrete Math., 19 (2005) 610-621.
[8] Daphne Der-Fen Liu, Radio Number for Trees, preprint 2006 (to appear)
[9] J. A. Gallian, A dynamic survey of graph labeling, The Electronic Journal of Combinatorics,
# DS6,(2009),1-219.
 Radio Number of Cube of a Path 29

[10] W.K. Hale, Frequency assignment: theory and applications, Proc. IEEE 68 (1980), 1497-
1514.
[11] Hartsfield Gerhard and Ringel, Pearls in Graph Theory, Academic Press, USA, 1994.
[12] B. Sooryanarayana and Raghunath.P, Radio labeling of cube of a cycle, Far East J. Appl.
Math 29(1)(2007), 113-147.
[13] B. Sooryanarayana and Raghunath.P, Radio labeling of Cn4 , Journal of Applied Mathemat-
ical Analysis and Applications 3(2007) no. 2, 195-228.
[14] J.Van den Heuvel, R.Leese and M.Shepherd, Graph labeling and radio Channel Assignment,
J.Graph Theory, 29(1998), 263-283.
[15] P.Zhang, Radio labellings of Cycles, Ars Combin., 65 (2002), 21-32.
Math.Combin.Book Ser. Vol.1 (2010), 30-41

Biprimitive Semisymmetric Graphs on P SL(2, p)

Furong Wang
(Beijing Wuzi University, Beijing 101149, P.R.China)

Lin Zhang

(Capital University of Economics and Business, Beijing 100070, P.R.China)

Email: [email protected], [email protected]

Abstract: A simple undirected graph is said to be semisymmetric if it is regular and

edge-transitive but not vertex-transitive. It is easy to see that every semisymmetric graph
is necessarily bipartite, with the two parts having equal size and the automorphism group
acting transitively on each of these two parts. A semisymmetric graph is called biprimitive
if its automorphism group acts primitively on each part. This paper gives a classification of
biprimitive semisymmetric graphs arising from the action of the group P SL(2, p) on cosets of
A5 , where p ≡ 1 (mod10) is a prime. By the way, the structure of the suborbits of P GL(2, p)
on the cosets of A5 is determined.

Keywords: Smarandache multi-group, group, semisymmetric graph, Biprimitive semisym-

metric graph, suborbit.

AMS(2010): 05C25, 20B25

§1. Introduction

For the group- and graph-theoretic terminology we refer the reader to [1,7]. All graphs con-
sidered in this paper are finite, undirected and simple. For a graph X, we use V (X), E(X),
A(X) and Aut(X) to denote its vertex set, edge set, arc set and full automorphism group,
respectively. If X be a bipartite with bipartition V (X) = U (X) ∪ W (X). Set

A+ = hg ∈ A | U (X)g = U (X), W (X)g = W (X)i .

Clearly, if X is connected then either |A : A+ | = 2 or A = A+ , depending on whether or not

there exists an automorphism which interchanges the two parts U (X) and W (X). Suppose G
is a subgroup of A+ . Then X is said to be G-semitransitive if G acts transitively on both U (X)
and W (X), and semitransitive if X is A+ -semitransitive. Also X is said to be biprimitive if A+
acts primitively on each part. We call a graph semisymmetric if it is regular and edge-transitive
but not vertex-transitive. It is easy to see that every semisymmetric graph is a bipartite graph
with two parts of equal size and is semitransitive.
The first person who studied semisymmetric graphs was Folkman. In 1967 he constructed
1 Supported by NNSF(10971144) and BNSF(1092010).
2 Received January 13, 2010. Accepted March 21, 2010.
 Biprimitive Semisymmetric Graphs on P SL(2, p) 31

several infinite families of such graphs and proposed eight open problems (see [6]). Afterwards,
Bouwer, Titov, Klin, A.V. Ivanov, A.A. Ivanov and others did much work on semisymmetric
graphs (see [2-3,8-10,13]). They gave new constructions of such graphs and nearly solved all of
Folkman’s open problems. In particular, Iofinova and Ivanov [9] in 1985 classified biprimitive
semisymmetric cubic graphs using group-theoretical methods; this was the first classification
theorem for such graphs. More recently, following some deep results in group theory which
depend on the classification of finite simple groups, some new methods and results in vertex-
transitive graphs and semisymmetric graphs have appeared. In [5], for example, the authors
give a classification of semisymmetric graphs of order 2pq where p and q are distinct primes. It
is shown that there are 131 examples of such graphs, which are biprimitive. In [4] a classification
is given, of biprimitive semisymmetric graphs arising from the action of the group P SL(2, p),
p ≡ 1 (mod8) a prime, on cosets of S4 . In this paper, we will classify all biprimitve graphs
arising from the action of the group P SL(2, p), p ≡ 1 (mod10) a prime, on cosets of A5 . To
prove the classification theorem, we have to determine the suborbits of P GL(2, p) acts on the
cosets of A5 and such a determination will certainly be useful for other problems.
Throughout the paper, Zn and Dn denote the cyclic group of order n and the dihedral
group of order n, respectively. A semidirect product of the group N by the group H will be
denoted by N : H. Given a group G and a subgroup H of G, we use [G : H] to denote the
set of right cosets of H in G. The action of G on [G : H] is always assumed to be the right
multiplication action. More precisely, for g ∈ G, we use R(g) to denote the effect of right
multiplication of g on [G : H] and let R(G) = {R(g)|g ∈ G}. However, for convenience, in most
cases we will identify R(g) with g, except for the special cases to be stated.
A Smarandache multi-group G is an union of groups (G1 ; ◦1 ), (G2 ; ◦2 ), · · · , (Gn ; ◦n ), dif-
ferent two by two for an integer n ≥ 1. Particularly, if n = 1, then G is just a group. A
Smarandache multi-group G is naturally acting on its underlying graph G[G ]. In [5], the au-
thors gave a group-theoretic construction of semitransitive graphs by introducing the definition
of so called bi-coset graph as following: Let G be a group, let L and R be subgroups of G
and let D be a union of double cosets of R and L in G, namely, D = ∪i Rdi L. Define a
bipartite graph X = B(G, L, R; D) with bipartition V (X) = [G : L] ∪ [G : R] and edge set
E(X) = {{Lg, Rdg}|g ∈ G, d ∈ D}. This graph is called the bi-coset graph of G with respect
to L, R and D.
Note that in the above construction of semitransitive graphs, if L and R are the same
subgroup, then we still use Lg and Rg to denote different vertices in the two parts of V (X).
It is proved in [5] that (1) the graph X = B(G, L, R; D) is a well-defined bipartite graph, and
under the right multiplication action on V (X) of G, the graph X is G-semitransitive; (2) every
G-semitranstive graph is isomorphic to one of such bi-coset graphs.
Now we state the main theorem of this paper.

Theorem 1.1 Let p ≡ 1(mod10), G = P SL(2, p) and Q = P GL(2, p). Let Y be a biprimitive
semisymmetric graph with a subgroup G of Aut(Y ) acting edge-transitively on Y and having
A5 as a vertex stabilizer. Then Y is isomorphic to one of the following graphs:

(i) B(G, L, L; D), where L ∼

= A5 and D is a double coset corresponds to a non-self-paired
32 Furong Wang and Lin Zhang

suborbit of G relative to L.
(ii) B(G, L, Lσ ; D), where σ is an involution in Q \ G and LσdL corresponds to a non-self-
paired suborbit of Q relative to L.
3
Moreover, each such graph Y is of order p 60 −p
and valency 60, and with the automorphism
group P SL(2, p). Table 1 lists the total numbers n1 and n2 of nonisomorphic semisymmetric
graphs B(G, L, L; D) and B(G, L, Lσ , D) for each of the congruence classes of p.

TABLE 1.
p(mod120) n1 n2
p3 −60p2 +1077p−15418 p3 −60p2 +1197p−1138
1 14400 14400
p3 −60p2 +1197p−13142 p3 −60p2 +1077p+1138
−1 14400 14400
p3 −60p2 +1197p−7238 p3 −60p2 +1077p−5918
11 14400 14400
p3 −60p2 +1077p−8362 p3 −60p2 +1197p−7024
−11 14400 14400
p3 −60p2 +1197p−9238 p3 −60p2 +1077p−5518
31 14400 14400
p3 −60p2 +1077p−8762 p3 −60p2 +1197p−5042
−31 14400 14400
p3 −60p2 +1077p−12218 p3 −60p2 +1197p−2738
41 14400 14400
p3 −60p2 +1197p−11542 p3 −60p2 +1077p−2062
−41 14400 14400
p3 −60p2 +1077p−11818 p3 −60p2 +1197p−4738
61 14400 14400
p3 −60p2 +1197p−9542 p3 −60p2 +1077p−2462
−61 14400 14400
p3 −60p2 +1197p−10838 p3 −60p2 +1077p−2318
71 14400 14400
p3 −60p2 +1077p−11962 p3 −60p2 +1197p−3442
−71 14400 14400
p3 −60p2 +1197p−5638 p3 −60p2 +1077p−9118
91 14400 14400
p3 −60p2 +1077p−5162 p3 −60p2 +1197p−8642
−91 14400 14400
p3 −60p2 +1077p−8618 p3 −60p2 +1197p−6338
101 14400 14400
p3 −60p2 +1197p−7942 p3 −60p2 +1077p−5662
−101 14400 14400

§2. Preliminaries

In this section, some preliminary results are given. The first two propositions give some prop-
erties of the groups P SL(2, p) and P GL(2, p).

Proposition 2.1 ([11], Lemma 2.1) Let p be an odd prime. Then

(1) the maximal subgroups of P SL(2, p) are:

One class of subgroups isomorphic to Zp : Z p−1 ; one class isomorphic to Dp−1 , when
2
p > 13; one class isomorphic to Dp+1 , when p 6= 7; two classes isomorphic to A5 , when p ≡
1(mod10); two classes isomorphic to S4 , when p ≡ 1(mod8); and one class isomorphic to A4 ,
when p = 5 or p 6≡ 1(mod8).
(2)The maximal subgroups of P GL(2, p) are:

One class of subgroups isomorphic to Zp : Zp−1 ; one class isomorphic to D2(p−1) , when
 Biprimitive Semisymmetric Graphs on P SL(2, p) 33

p > 7; one class isomorphic to D2(p+1) ; one class isomorphic to S4 , when p = 5 or p 6≡ 1(mod40)
and p > 5; and one subgroup P SL(2, p).

Proposition 2.2 ([5], Lemma 3.9) Any extension of P SL(2, p) by Z2 is isomorphic to either
P GL(2, p) or P SL(2, p) × Z2 . In both cases the extension is split.

Proposition 2.3 ([5], Lemma 2.3) The graph X = B(G, L, R; D) is a well-defined bipartite
graph. Under the right multiplication action on V (X) of G, the graph X is G-semitransitive.
The kernel of the action of G on V (X) is CoreG (L) ∩ CoreG (R), the intersection of the cores
of the subgroups L and R in G. Furthermore, we have
(i) X is G-edge-transitive if and only if D = RdL for some d ∈ G;
(ii) the degree of any vertex in [G : L] (resp. [G : R]) is equal to the number of right cosets
of R (resp. L) in D (resp. D−1 ), so X is regular if and only if |L| = |R|;
(iii) X is connected if and only if G is generated by elements of D−1 D;
S
(iv) X ∼= B(G, La , Rb ; D′ ) where D′ = i Rb (b−1 di a)La , for any a, b ∈ G.
The next proposition provides one general and three particular conditions, each of which
is sufficient for a G-semitransitive graph to be vertex-transitive.

Proposition 2.4 ([5], Lemma 2.6) Let X = B(G, L, R; D). If there exists an involutory auto-
morphism σ of G such that Lσ = R and Dσ = D−1 , then X is vertex-transitive. In particular,
(i) If G is abelian and acts regularly on both parts of X, then X is vertex-transitive. In
other words, bi-Cayley graphs of abelian groups are vertex-transitive.
(ii) If there exists an involutory automorphism σ of G such that Lσ = R, and the lengths
of the orbits of L on [G : R] (or the orbits of R on [G : L]) are all distinct, then X is vertex-
transitive.
(iii) If the representations of G on the two parts of X are equivalent and all suborbits of G
relative to L are self-paired, then X is vertex-transitive.

The link between groups and graphs that we use is the concept of the orbital graph of a
permutation group. For the terminology of orbital graph we refer the reader to [12].
The following group theoretical results will be used later.

Proposition 2.5 ([11], Lemma 2.1) Let G be a transitive group on Ω and let H = Gα for some
α ∈ Ω. Suppose that K 6 G and at least one G-conjugate of K is contained in H. Suppose
further that the set of G-conjugates of K which are contained in H form t conjugacy classes of
Pt
H with representatives K1 , K2 , · · · , Kt . Then K fixes i=1 |NG (Ki ) : NH (Ki )| points of Ω.

Proposition 2.6 ([11], Lemma 2.2) Let G be a primitive permutation group on Ω, and let
H = Gα for some α ∈ Ω. Suppose that H = A5 and let K1 , ... , K7 be seven subgroups of H
satisfying K1 ∼ = A4 , K2 ∼= D10 , K3 ∼= D6 , K 4 ∼ = Z3 , K 6 ∼
= Z5 , K 5 ∼ = D4 and K7 ∼ = Z2 . Let ki
be the number of points in Ω fixed by Ki , for i = 1, 2, ... , 7. Then G has 1 suborbit of length 1,
k1 − 1 suborbits of length 5, k2 − 1 suborbits of length 6, k3 − 1 suborbits of length 10, 12 (k4 − k2 )
suborbits of length 12, 12 (k5 − 2k1 − k3 + 2) suborbits of length 20, 31 (k6 − k1 ) suborbits of length
15, 12 (k7 − 2k2 − 2k3 − k6 + 4) suborbits of length 30, and all the other suborbits have length 60.

Proposition 2.7 ([11], Lemma 2.3) Let D = D2n be the dihedral group of order 2n, considered
34 Furong Wang and Lin Zhang

as a permutation group of degree n generated by a = (1, 2, · · · , n) and b = (1)(2, n)(3, n −

1) · · · (i, n + 2 − i) · · · , for any n > 2. Then the nontrival orbitals of D are Γi = (1, i)D =
(1, n + 2 − i)D , for 2 6 i 6 (n + 2)/2. Each of these orbitals is self-paired. Moreover, for all
points i, j, with i 6= j, there is an involution in D which interchanges i and j.

Proposition 2.8 ([11], Lemma 2.4) Let G be a transitive group on Ω and let H = Gα for
some α ∈ Ω. Suppose that G has t conjugacy classes of involutions, say C1 , · · · , Ct . Suppose
further that a representative uj in Cj has Nj cycles of length 2, and that the centralizer of
uj in G has order cj . Also for a nontrivial self-paired suborbit ∆ relative to α and a point
P N
B ∈ ∆, let inv(∆) be the number of involutions in G with a 2-cycle (σ, B). Then tj=1 cjj =
1 P
2|H| ∆=∆∗ |∆(α)|inv(∆), where cj is the order of the centralizer of uj .

§3. Proof of Theorem 1.1

Now we begin the proof of Theorem 1.1. From now on we shall assume that G = P SL(2, p) and
Q = P GL(2, p), where p ≡ 1 (mod10). Clearly, Q = G : hσi for some involution σ ∈ Q \ G. Let
Y be a semisymmetric biprimitive graph with a subgroup G of Aut(Y ) acting edge-transitively
on Y and having A5 as a vertex stabilizer. Let U (Y ) and W (Y ) be the bipartition of V (Y ).
3
Then |U (Y )| = |W (Y )| = p120−p
and Gv ∼
= A5 for any v ∈ U (Y ) and v ∈ W (Y ). Now Y is
isomorphic to the bi-coset graph X = B(G, L, R; D), where L ∼ =R∼ = A5 . With our notation,
V (X) = U (X) ∪ W (X) = [G : L] ∪ [G : R]. We will treat the following two cases separately:

(1) Suppose the representations of G on U (X) and W (X) are equivalent. In this case,
by Proposition 2.3 (iv), no loss of any generality, we may assume L = R ∼ = A5 . With the
completely similar arguments as in [5, Lemma 4.1], we may show that X is semisymmetric if
and only if D−1 6= D, that is, D corresponds to a non-self-paired suborbit of G relative to L,
and two such bi-coset graphs defined (for the same group G) by distinct double cosets D1 and
D2 are isomorphic if and only if D1 and D2 are paired with each other in G, or more precisely,
D1 = D2−1 .
(2) Suppose the representations of G on U (X) and W (X) are inequivalent. Let Q =
P GL(2, p) = hG, σi, where σ ∈ Q \ G and σ 2 = 1. By the Proposition 2.1, G has two conjugacy
classes of subgroups isomorphic to A5 , which are fused by σ. Therefore, we may let R = Lσ
so that X = B(G, L, Lσ ; D) where D = Lσ dL for some d ∈ G. With the similar arguments as
in [5, Lemma 4.2], X is semisymmetric if and only if the suborbit LσdL of Q relative to L is
not self-paired, and two such graphs X1 = B(G, L, R; D1 ) and X2 = B(G, L, R; D2 ) defined by
distinct double cosets D1 := Rd1 L and D2 := Rd2 L respectively are isomorphic if and only if
D1′ := Lσd1 L and D2′ := Lσd2 L are paired with each other in Q = P GL(2, p).
Following the above two cases, we need to determine non-self-paired suborbits of G relative
to L and non-self-paired suborbits of Q relative to L which are contained in [Q : L] \ [G : L].
Noting that the number of non-self-paired suborbits of G relative to L is the same as the number
of non-self-paired suborbits of Q relative to L which are contained in [G : L]. From now on
let Ω = [Q : L], Ω1 = [G : L] and Ω2 = [Q : L] \ [G : L]. We will consider the action of Q on
Ω and find all non-self-paired suborbits of Q contained in Ω1 and in Ω2 as well. We shall do
this only for the case where G = P SL(2, p) and L = A5 , p ≡ 1(mod 120), and for the other
 Biprimitive Semisymmetric Graphs on P SL(2, p) 35

cases, similar arguments and computations lead to the data listed in Appendix: TABLE 4 − 1-
TABLE 4 − 4.
Let Ki (for 1 6 i 6 7) be the representatives of the seven conjugacy classes of nontrivial
subgroups of L isomorphic to A4 , D10 , D6 , Z5 , Z3 , D4 and Z2 , respectively, and let K8 = 1.
Clearly any nontrivial subgroup K of L with a fixed point on Ω must be conjugate to one of
these Ki . For each i ∈ {1, . . . , 8}, let ki , ki1 and ki2 denote the respective numbers of fixed
points of Ki in Ω, Ω1 and Ω2 . Among of all the suborbits with the L-stabilizer Ki , let xi1
and xi2 denote the respective numbers of the suborbits contained in Ω1 and Ω2 ; let yi , yi1 ,
yi2 = yi − yi1 denote the respective numbers of self-paired suborbits contained in Ω, Ω1 and
Ω2 ; and let hi1 = xi1 − yi1 and hi2 = xi2 − yi2 denote the respective numbers of non-self-paired
suborbits contained in Ω1 and Ω2 .
First we determine the values of xi1 and xi2 . For i ∈ {1, . . . , 7}, these values are given
in TABLE 2 and are obtained in the following way. After having determined the respective
normalizers of each Ki in L and in G (resp. Q), we apply Proposition 2.5 to calculate ki1 (resp.
ki ). Then ki2 = ki − ki1 can be found also. By Proposition 2.6, we can determine the values of
xi1 and xi2 , 1 ≤ i ≤ 7.
TABLE 2.
i 1 2 3 4 5 6 7
Ki A4 D10 D6 Z5 Z3 D4 Z2
NL (Ki ) A4 D10 D6 D10 D6 A4 D4
NG (Ki ) S4 D20 D12 Dp−1 Dp−1 S4 Dp−1
p−1 p−1 p−1
ki1 2 2 2 10 6
2 4
p−1 p−1 p−1
xi1 1 1 1 20
−1 12
−2 0 8
−3
NQ (Ki ) S4 D20 D12 D2(p−1) D2(p−1) S4 D2(p−1)
p−1 p−1 p−1
ki 2 2 2 5 3
2 2
p−1 p−1 p−1
ki2 0 0 0 10
−1 6
0 4
p−1 p−1 p−1
xi2 0 0 0 20 12
0 8

Finally,
7
!
1 p3 − p X 60
x81 = −1− xi1
60 120 i=1
|Ki |
 3  
1 p −p p−1
= − 1 − 1 · 5 − 1 · 6 − 1 · 10 − − 1 · 12
60 120 20
    
p−1 p − 1
− − 2 · 20 − − 3 · 30
12 8
p3 − 723p + 15122
=
7200
and a similar computation gives
7
!
1 p3 − p X 60
x82 = −1− xi2
60 120 i=2
|K i|
36 Furong Wang and Lin Zhang

 
1 p3 − p p−1 p−1 p−1
= −1− · 12 − · 20 − · 30
60 120 20 12 8
p3 − 723p + 722
= .
7200
Next we determine the values of hi1 and hi2 . We claim all the non-regular suborbits of Q are
self-paired, so that hi1 = hi2 = 0 for 1 6 i 6 7. For example, let i = 7 and let ∆ be a suborbit
with L-stabilizer K7 = Z2 , and take v ∈ ∆. We consider the action of NQ (K7 ) ∼ = D2(p−1) on
F ix(K7 ), the set of fixed points of K7 on Ω. This action is transitive and the kernel is Z2 .
Since |F ix(K7 )| = p−1
2 , by Proposition 2.7, there exists an element in NQ (K7 ) interchanging
u = L and v. So ∆ is self-paired, or equivalently, h71 = h72 = 0.
It remains to determine h81 and h82 , the numbers of non-self-paired suborbits of Q in Ω1
and in Ω2 respectively. For these it suffices to calculate y81 and y8 , the numbers of self-paired
regular suborbits of Q in Ω1 and in Ω, since h81 = x81 − y81 , h82 = x82 − y82 and y8 = y81 + y82 .
By Proposition 2.8, in order to calculate y81 (resp. y8 ), we need the value of inv(∆), which
is defined in Proposition 2.8 for all self-paired suborbits ∆ of G (resp. Q). Furthermore, to
calculate inv(∆) we need to know Guv and G{u,v} (resp. Quv and Q{u,v} ), where u = L and
v ∈ ∆.
The lengths li (1 6 i 6 8) of self-paired suborbits with point stabilizer Ki , the numbers
yi1 and yi , the groups Guv , G{u,v} and Quv and Q{u,v} , and the value of inv(∆) for each ∆ are
listed in the following table.

TABLE 3.
i li yi1 yi Guv = Quv G{u,v} = Q{u,v} inv(∆)
1 5 1 1 A4 S4 6
2 6 1 1 D10 D20 6
3 10 1 1 D6 D12 4
p−1 p−1
4 12 20
−1 10
−1 Z5 D10 5
p−1 p−1
5 20 12
−2 6
−2 Z3 D6 3
p−1 p−1
7 30 8
−3 4
−3 Z2 D4 2
8 60 y81 y8 1 Z2 1

Next we shall calculate y81 and y8 using Proposition 2.8. We know that Q has two conjugacy
classes of involutions. A representative of the first class, say u1 ∈ G, fixes p−1
2 points, and so u1
p3 −p p−1
3
−
contains N1 = 60 2 2 = p −31p+30120 cycles of length 2. Further, CQ (u1 ) ∼
= D2(p−1) has order
c1 = 2(p − 1). A representative of the second class, say u2 ∈ Q \ G, has no fixed point and so
3
u2 contains N2 = p120−p
cycles of length 2. Also CQ (u2 ) ∼
= D2(p+1) has order c2 = 2(p + 1). By
Proposition 2.8 and TABLE 3, we have

p3 − 31p + 30 p3 − p 1
+ = (1 · 5 · 6 + 1 · 6 · 6 + 1 · 10 · 4
240(p − 1) 240(p + 1) 2 · 60
      
p−1 p−1 p−1
+ − 1 · 12 · 5 + − 2 · 20 · 3 + − 3 · 30 · 2 + 60y8 .
10 6 4
p2 −31p+270
It follows that y8 = 60 .
 Biprimitive Semisymmetric Graphs on P SL(2, p) 37

To determine y81 and y82 , we turn to the group G. Note that G has only one conjugacy
class of involutions, and each involution u has precisely p−1
4 fixed points in Ω1 and so has
p3 −p p−1 3
−
N = 120 2 4 = p −31p+30 240 cycles of length 2. Also CG (u) ∼ = Dp−1 has order c = p −
2
1. By Proposition 2.8 and TABLE 3, we may calculcate y81 = p −30p+509 120 . Hence y82 =
2 3 2
y8 − y81 = p −32p+31
120 and so h 81 = x81 − y 81 = p −60p +1077p−15418
7200 and h 82 = x82 − y82 =
p3 −60p2 +1197p−1138
7200 .
3 2
Hence we find that Q has p −60p +1077p−15418
7200 non-self-paired regular suborbits, which
p3 −60p2 +1197p−1138
have length 60 and are contained in Ω1 and Q has 7200 non-self-paired regular
3 2
suborbits, which have length 60 and are contained in Ω2 . So we have p −60p +1077p−15418 14400
3 2
semisymmetric graphs X with valency 60 in case (i) and p −60p 14400 +1197p−1138
semisymmetric
graphs X with valency 60 in case (ii) , as listed in TABLE 1.

Thus we finish the proof of Theorem 1.1. 

References

[1] N.L. Biggs, Algebraic graph theory, Cambridge University Press, 1974.
[2] I.Z. Bouwer, On edge but not vertex transitive cubic graphs, Canad. Math. Bull. 11(1968),
533-535.
[3] I.Z. Bouwer, On edge but not vertex transitive regular graphs, J. Combin. Theory Ser. B,
12(1972), 32-40.
[4] S.F. Du and Dragan Marus̆ic̆, An infinite family of biprimitive semisymmetric graphs, J.
Graph Theory, 32(1999), 217-228.
[5] S.F. Du and M.Y. Xu, A classification of semisymmetric graphs of order 2pq, Communi-
cations In Algbra, 28(2000), 2685-2715.
[6] J. Folkman, Regular line-symmetric graphs, J. Combin. Theory Ser. B, 3(1967), 215-232.
[7] B. Huppert, Endliche Gruppen I, Springer-Verlag, 1967.
[8] A.V. Ivanov, On edge but not vertex transitive regular graphs, Comb. Annals of Discrete
Mathematices, 34(1987), 273-286.
[9] M.E. Iofinova and A.A. Ivanov, Biprimitive cubic graphs (Russian), in Investigation in Al-
gebraic Theory of Combinatorial Objects, Proceedings of the seminar, Institute for System
Studies, Moscow, 1985, pp. 124-134.
[10] M.H. Klin, On edge but not vertex transitive regular graphs, Colloquia Mathematica Soci-
etatis Janos Bolyai, 25. Algebraic Methods in Graph Theory, Szeged (Hungary), Budapest,
1981, pp. 399-403.
[11] C.E. Praeger and M.Y. Xu, Vertex primitive transitive graphs of order a product of two
distinct primes, J. Combin. Theory Ser. B, 59(1993), 245-266.
[12] D.E. Taylor and M.Y. Xu, Vertex-primitive 1/2-transitive graphs, J. Austral. Math. Soc.
Ser. A, 57(1994), 113-124.
[13] V.K. Titov, On symmetry in the graphs (Russian), Voprocy Kibernetiki (15), Proceedings
of the II All Union Seminar on Combinatorial Mathematices, Part 2, Nauka, Moscow,
1975, pp. 76-109.
38 Furong Wang and Lin Zhang

Appendix:

TABLE 4-1
i 1 2 3 4 5 7 8
i 1 2 3 4 5 7 8
Ki A4 D10 D6 Z5 Z3 Z2 1
NL (Ki ) A4 D10 D6 D10 D6 D4
NG (Ki ) S4 D20 D12 Dp+1 Dp+1 Dp+1
p+1 p+1 p+1
ki1 2 2 2 10 6 4
p+1 p+1 p+1 p3 −723p+13678
xi1 1 1 1 20 − 1 12 − 2 8 − 3 7200
p+1 p+1 p+1 p2 −32p+447
yi1 1 1 1 20 − 1 12 − 2 8 − 3 120
p3 −60p2 +1197p−13142
hi1 0 0 0 0 0 0 7200
p ≡ −1(mod120) NQ (Ki ) S4 D20 D12 D2(p+1) D2(p+1) D2(p+1)
p+1 p+1 p+1
ki 2 2 2 5 3 2
p+1 p+1 p+1
ki1 0 0 0 10 6 4
p+1 p+1 p+1 p3 −723p−722
xi2 0 0 0 20 12 8 7200
p+1 p+1 p+1 p2 −30p−31
yi2 0 0 0 20 12 8 120
p3 −60p2 +1077p+1138
hi2 0 0 0 0 0 0 7200
NG (Ki ) A4 D10 D12 Dp−1 Dp+1 Dp+1
p−1 p+1 p+1
ki1 1 1 2 10 6 4
p−1 1 p+1 p+1 3 p3 −723p+6622
xi1 0 0 1 20 − 2 12 − 1 8 − 2 7200
p−1 1 p+1 p+1 3 p2 −32p+231
yi1 0 0 1 20 − 2 12 − 1 8 − 2 120
p3 −60p2 +1197p−7238
hi1 0 0 0 0 0 0 7200
p ≡ 11(mod120) NQ (Ki ) S4 D20 D12 D2(p−1) D2(p+1) D2(p+1)
p−1 p+1 p+1
ki 2 2 2 5 3 2
p−1 p+1 p+1
ki2 1 1 0 10 6 4
p−1 1 p+1 p+1 3 p3 −723p+6622
xi2 1 1 0 20 − 2 12 − 1 8 − 2 7200
p−1 1 p+1 p+1 3 p2 −30p+209
yi2 1 1 0 20 − 2 12 − 1 8 − 2 120
p3 −60p2 +1077p−5918
hi2 0 0 0 0 0 0 7200
NG (Ki ) A4 D10 D12 Dp+1 Dp−1 Dp−1
p+1 p−1 p−1
ki1 1 1 2 10 6 4
p+1 1 p−1 p−1 3 p3 −723p+7778
xi1 0 0 1 20 − 2 12 − 1 8 − 2 7200
p+1 1 p−1 p−1 3 p2 −32p+269
yi1 0 0 1 20 − 2 12 − 1 8 − 2 120
p3 −60p2 +1077p−8362
hi1 0 0 0 0 0 0 7200
p ≡ −11(mod120) NQ (Ki ) S4 D20 D12 D2(p+1) D2(p−1) D2(p−1)
p+1 p−1 p−1
ki 2 2 2 5 3 2
p+1 p−1 p−1
ki2 1 1 0 10 6 4
p+1 1 p−1 p−1 3 p3 −723p+7778
xi2 1 1 0 20 − 2 12 − 1 8 − 2 7200
p+1 1 p−1 p−1 3 p2 −32p+247
yi2 1 1 0 20 − 2 12 − 1 8 − 2 120
p3 −60p2 +1197p−7042
hi2 0 0 0 0 0 0 7200
 Biprimitive Semisymmetric Graphs on P SL(2, p) 39

TABLE 4-2
i 1 2 3 4 5 7 8
Ki A4 D10 D6 Z5 Z3 Z2 1
NL (Ki ) A4 D10 D6 D10 D6 D4
NG (Ki ) S4 D10 D6 Dp−1 Dp−1 Dp+1
p−1 p−1 p+1
ki1 2 1 1 10 6 4
p−1 1 p−1 3 p+1 p3 −723p+7022
xi1 1 0 0 20 − 2 12 − 2 8 − 1 7200
p−1 1 p−1 3 p+1 p2 −32p+271
yi1 1 0 0 20 − 2 12 − 2 8 − 1 120
p3 −60p2 +1197p−9238
hi1 0 0 0 0 0 0 7200
p ≡ 31(mod120) NQ (Ki ) S4 D20 D12 D2(p−1) D2(p−1) D2(p+1)
p−1 p−1 p+1
ki 2 2 2 5 3 2
p−1 p−1 p+1
ki2 0 1 1 10 6 4
p−1 1 p−1 1 p+1 p3 −723p+7022
xi2 0 1 1 20 − 2 12 − 2 8 − 2 7200
p−1 1 p−1 1 p+1 p2 −30p+209
yi2 0 1 1 20 − 2 12 − 2 8 − 2 120
p3 −60p2 +1077p−5518
hi2 0 0 0 0 0 0 7200
NG (Ki ) S4 D10 D6 Dp+1 Dp+1 Dp−1
p+1 p+1 p−1
ki1 2 1 1 10 6 4
p+1 1 p+1 3 p−1 p3 −723p+7378
xi1 1 0 0 20 − 2 12 − 2 8 − 1 7200
p+1 1 p+1 3 p−1 p2 −30p+269
yi1 1 0 0 20 − 2 12 − 2 8 − 1 120
p3 −60p2 +1077p−8762
hi1 0 0 0 0 0 0 7200
p ≡ −31(mod120) NQ (Ki ) S4 D20 D12 D2(p+1) D2(p+1) D2(p−1)
p+1 p+1 p−1
ki 2 2 2 5 3 2
p+1 p+1 p−1
ki2 0 1 1 10 6 4
p+1 1 p+1 1 p−1 p3 −723p+7378
xi2 0 1 1 20 − 2 12 − 2 8 − 2 7200
p+1 1 p+1 1 p−1 p2 −32p+207
yi2 0 1 1 20 − 2 12 − 2 8 − 2 120
p3 −60p2 +1197p−5042
hi2 0 0 0 0 0 0 7200
NG (Ki ) S4 D20 D6 Dp−1 Dp+1 Dp−1
p−1 p+1 p−1
ki1 2 2 1 10 6 4
p−1 p+1 3 p−1 p3 −723p+11122
xi1 1 1 0 20 − 1 12 − 2 8 − 2 7200
p−1 p+1 3 p−1 p2 −30p+389
yi1 1 1 0 20 − 1 12 − 2 8 − 2 120
p3 −60p2 +1077p−12218
hi1 0 0 0 0 0 0 7200
p ≡ 41(mod120) NQ (Ki ) S4 D20 D12 D2(p−1) D2(p+1) D2(p−1)
p−1 p+1 p−1
ki 2 2 2 5 3 2
p−1 p+1 p−1
ki2 0 0 1 10 6 4
p−1 p+1 1 p−1 p3 −723p+3922
xi2 0 0 1 20 12 − 2 8 − 1 7200
p−1 p+1 1 p−1 p2 −32p+111
yi2 0 0 1 20 12 − 2 8 − 1 120
p3 −60p2 +1197p−2738
hi2 0 0 0 0 0 0 7200
NG (Ki ) S4 D20 D6 Dp+1 Dp−1 Dp+1
p+1 p−1 p+1
ki1 2 2 1 10 6 4
p+1 p−1 3 p+1 p3 −723p+10478
xi1 1 1 0 20 − 1 12 − 2 8 − 2 7200
p+1 p−1 3 p+1 p2 −32p+367
yi1 1 1 0 20 − 1 12 − 2 8 − 2 120
p3 −60p2 +1197p−11542
hi1 0 0 0 0 0 0 7200
p ≡ −41(mod120) NQ (Ki ) S4 D20 D12 D2(p+1) D2(p−1) D2(p+1)
p+1 p−1 p+1
ki 2 2 2 5 3 2
p+1 p−1 p+1
ki2 0 0 1 10 6 4
p+1 p−1 1 p+1 p3 −723p+3278
xi2 0 0 1 20 12 − 2 8 − 1 7200
p+1 p−1 1 p+1 p2 −30p+89
yi2 0 0 1 20 12 − 2 8 − 1 120
p3 −60p2 +1077p−2062
hi2 0 0 0 0 0 0 7200
40 Furong Wang and Lin Zhang

TABLE 4-3
i 1 2 3 4 5 7 8
i 1 2 3 4 5 7 8
Ki A4 D10 D6 Z5 Z3 Z2 1
NL (Ki ) A4 D10 D6 D10 D6 D4
NG (Ki ) A4 D20 D12 Dp−1 Dp−1 Dp−1
p−1 p−1 p−1
ki1 1 2 2 10 6 4
p−1 p−1 p−1 5 p3 −723p+11522
xi1 0 1 1 20 − 1 12 − 1 8 − 2 7200
p−1 p−1 p−1 5 p2 −30p+389
yi1 0 1 1 20 − 1 12 − 1 8 − 2 120
p3 −60p2 +1077p−11818
hi1 0 0 0 0 0 0 7200
p ≡ 61(mod120) NQ (Ki ) S4 D20 D12 D2(p−1) D2(p−1) D2(p−1)
p−1 p−1 p−1
ki 2 2 2 5 3 2
p−1 p−1 p−1
ki2 1 0 0 10 6 4
p−1 p−1 p−1 1 p3 −723p+4322
xi2 1 0 0 20 12 − 1 8 − 2 7200
p−1 p−1 p−1 1 p2 −32p+151
yi2 1 0 0 20 12 − 1 8 − 2 120
p3 −60p2 +1197p−4738
hi2 0 0 0 0 0 0 7200
NG (Ki ) A4 D20 D12 Dp+1 Dp+1 Dp+1
p+1 p+1 p+1
ki1 1 2 2 10 6 4
p+1 p+1 p+1 5 p3 −723p+10078
xi1 0 1 1 20 − 1 12 − 1 8 − 2 7200
p+1 p+1 p+1 5 p2 −32p+327
yi1 0 1 1 20 − 1 12 − 1 8 − 2 120
p3 −60p2 +1197p−9542
hi1 0 0 0 0 0 0 7200
p ≡ −61(mod120) NQ (Ki ) S4 D20 D12 D2(p+1) D2(p+1) D2(p+1)
p+1 p+1 p+1
ki 2 2 2 5 3 2
p+1 p+1 p+1
ki2 1 0 0 10 6 4
p+1 p+1 p+1 1 p3 −723p+2878
xi2 1 0 0 20 12 − 1 8 − 2 7200
p+1 p+1 p+1 1 p2 −30p+89
yi2 1 0 0 20 12 − 1 8 − 2 120
p3 −60p2 +1077p−2462
hi2 0 0 0 0 0 0 7200
NG (Ki ) S4 D10 D12 Dp−1 Dp+1 Dp+1
p−1 p+1 p+1
ki1 2 1 2 10 6 4
p−1 1 p+1 p+1 p3 −723p+10222
xi1 1 0 1 20 − 2 12 − 2 8 − 2 7200
p−1 1 p+1 p+1 p2 −32p+351
yi1 1 0 1 20 − 2 12 − 2 8 − 2 120
p3 −60p2 +1197p−10838
hi1 0 0 0 0 0 0 7200
p ≡ 71(mod120) NQ (Ki ) S4 D20 D12 D2(p−1) D2(p+1) D2(p+1)
p−1 p+1 p+1
ki 2 2 2 5 3 2
p−1 p+1 p+1
ki2 0 1 0 10 6 4
p−1 1 p+1 p+1 p3 −723p+3022
xi2 0 1 0 20 − 2 12 8 − 1 7200
p−1 1 p+1 p+1 p2 −30p+89
yi2 0 1 0 20 − 2 12 8 − 1 120
p3 −60p2 +1077p−2318
hi2 0 0 0 0 0 0 7200
NG (Ki ) S4 D10 D12 Dp+1 Dp−1 Dp−1
p+1 p−1 p−1
ki1 2 1 2 10 6 4
p+1 1 p−1 p−1 p3 −723p+11378
xi1 1 0 1 20 − 2 12 − 2 8 − 2 7200
p+1 1 p−1 p−1 p2 −30p+389
yi1 1 0 1 20 − 2 12 − 2 8 − 2 120
p3 −60p2 +1077p−11962
hi1 0 0 0 0 0 0 7200
p ≡ −71(mod120) NQ (Ki ) S4 D20 D12 D2(p+1) D2(p−1) D2(p−1)
p+1 p−1 p−1
ki 2 2 2 5 3 2
p+1 p−1 p−1
ki2 0 1 0 10 6 4
p+1 1 p−1 p−1 p3 −723p+4178
xi2 0 1 0 20 − 2 12 8 − 1 7200
p+1 1 p−1 p−1 p2 −32p+127
yi2 0 1 0 20 − 2 12 8 − 1 120
p3 −60p2 +1197p−3442
hi2 0 0 0 0 0 0 7200
 Biprimitive Semisymmetric Graphs on P SL(2, p) 41

TABLE 4-4
i 1 2 3 4 5 7 8
Ki A4 D10 D6 Z5 Z3 Z2 1
NL (Ki ) A4 D10 D6 D10 D6 D4
NG (Ki ) A4 D10 D6 Dp−1 Dp−1 Dp+1
p−1 p−1 p+1
ki1 1 1 1 10 6 4
p−1 1 p−1 1 p+1 1 p3 −723p+3422
xi1 1 1 1 20 − 2 12 − 2 8 − 2 7200
p−1 1 p−1 1 p+1 1 p2 −32p+151
yi1 1 1 1 20 − 2 12 − 2 8 − 2 120
p3 −60p2 +1197p−5638
hi1 0 0 0 0 0 0 7200
p ≡ 91(mod120) NQ (Ki ) S4 D20 D12 D2(p−1) D2(p−1) D2(p+1)
p−1 p−1 p+1
ki 2 2 2 5 3 2
p−1 p−1 p+1
ki2 1 1 1 10 6 4
p−1 1 p−1 3 p+1 5 p3 −723p+10622
xi2 1 1 1 20 − 2 12 − 2 8 − 2 7200
p−1 1 p−1 3 p+1 5 p2 −30p+329
yi2 1 1 1 20 − 2 12 − 2 8 − 2 120
p3 −60p2 +1077p−9118
hi2 0 0 0 0 0 0 7200
NG (Ki ) A4 D10 D6 Dp+1 Dp+1 Dp−1
p−1 p+1 p−1
ki1 1 1 1 10 6 4
p+1 1 p+1 1 p−1 1 p3 −723p+3778
xi1 1 1 1 20 − 2 12 − 2 8 − 2 7200
p+1 1 p+1 1 p−1 1 p2 −30p+149
yi1 1 1 1 20 − 2 12 − 2 8 − 2 120
p3 −60p2 +1077p−5162
hi1 0 0 0 0 0 0 7200
p ≡ −91(mod120) NQ (Ki ) S4 D20 D12 D2(p+1) D2(p+1) D2(p−1)
p+1 p+1 p−1
ki 2 2 2 5 3 2
p+1 p+1 p−1
ki2 1 1 1 10 6 4
p+1 1 p+1 3 p−1 5 p3 −723p+10978
xi2 1 1 1 20 − 2 12 − 2 8 − 2 7200
p+1 1 p+1 3 p−1 5 p2 −32p+327
yi2 1 1 1 20 − 2 12 − 2 8 − 2 120
p3 −60p2 +1197p−8642
hi2 0 0 0 0 0 0 7200
NG (Ki ) A4 D20 D6 Dp−1 Dp+1 Dp−1
p−1 p+1 p−1
ki1 1 2 1 10 6 4
p−1 p+1 1 p−1 3 p3 −723p+7522
xi1 0 1 0 20 − 1 12 − 2 8 − 2 7200
p−1 p+1 1 p−1 3 p2 −30p+269
yi1 0 1 0 20 − 1 12 − 2 8 − 2 120
p3 −60p2 +1077p−8618
hi1 0 0 0 0 0 0 7200
p ≡ 101(mod120) NQ (Ki ) S4 D20 D12 D2(p−1) D2(p+1) D2(p−1)
p+1 p−1 p+1
ki 2 2 2 5 3 2
p−1 p+1 p−1
ki2 1 0 1 10 6 4
p−1 p+1 3 p−1 3 p3 −723p+7522
xi2 1 0 1 20 12 − 2 8 − 2 7200
p−1 p+1 3 p−1 3 p2 −32p+231
yi2 1 0 1 20 12 − 2 8 − 2 120
p3 −60p2 +1197p−6338
hi2 0 0 0 0 0 0 7200
NG (Ki ) A4 D20 D6 Dp+1 Dp−1 Dp+1
p+1 p−1 p+1
ki1 1 2 1 10 6 4
p+1 p−1 1 p+1 3 p3 −723p+6878
xi1 0 1 0 20 − 1 12 − 2 8 − 2 7200
p+1 p−1 1 p+1 3 p2 −32p+247
yi1 0 1 0 20 − 1 12 − 2 8 − 2 120
p3 −60p2 +1197p−7942
hi1 0 0 0 0 0 0 7200
p ≡ −101(mod120) NQ (Ki ) S4 D20 D12 D2(p+1) D2(p−1) D2(p+1)
p+1 p−1 p+1
ki 2 2 2 5 3 2
p+1 p−1 p+1
ki2 1 0 1 10 6 4
p+1 p−1 3 p+1 3 p3 −723p+6878
xi2 1 0 1 20 12 − 2 8 − 2 7200
p+1 p−1 3 p+1 3 p2 −30p+209
yi2 1 0 1 20 12 − 2 8 − 2 120
p3 −60p2 +1077p−5662
hi2 0 0 0 0 0 0 7200
Math.Combin.Book Ser. Vol.1 (2010), 42-46

A Note on Path Signed Digraphs

P. Siva Kota Reddy† S. Vijay‡ and H. C. Savithri∗

†
Department of Mathematics, Rajeev Institute of Technology, Industrial Area,

B-M Bypass Road, Hassan 573 201, India

‡
Department of Mathematics, Govt. First Grade College, Kadur, Chikkamangalore 577 548, India

∗
Department of Computer Science & Engineering, Rajeev Institute of Technology,

Industrial Area, B-M Bypass Road, Hassan 573 201, India

E-mail: reddy [email protected]

Abstract: A Smarandachely k-signed digraph (Smarandachely k-marked digraph) is an

ordered pair S = (D, σ) (S = (D, µ)) where D = (V, A) is a digraph called underlying
digraph of S and σ : A → (e1 , e2 , ..., ek ) (µ : V → (e1 , e2 , ..., ek )) is a function, where each
ei ∈ {+, −}. Particularly, a Smarandachely 2-signed digraph or Smarandachely 2-marked
digraph is called abbreviated a signed digraph or a marked digraph. In this paper, we define
−
→ −
→
the path signed digraph Pk (S) = (Pk (D), σ ′ ) of a given signed digraph S = (D, σ) and offer
a structural characterization of signed digraphs that are switching equivalent to their 3-path
−
→
signed digraphs P3 (S). The concept of a line signed digraph is generalized to that of a path
signed digraphs. Further, in this paper we discuss the structural characterization of path
−
→
signed digraphs Pk (S).

Key Words: Smarandachely k-Signed digraphs, Smarandachely k-marked digraphs, signed

digraphs, marked digraphs, balance, switching, path signed digraphs, line signed digraphs,
negation.

AMS(2000): 05C22

§1. Introduction
For standard terminology and notion in digraph theory, we refer the reader to the classic text-
books of Bondy and Murty [2]and Harary et al. [4]; the non-standard will be given in this paper
as and when required.
A Smarandachely k-signed digraph (Smarandachely k-marked digraph) is an ordered pair
S = (D, σ) (S = (D, µ)) where D = (V, A) is a digraph called underlying digraph of S and σ :
A → (e1 , e2 , ..., ek ) (µ : V → (e1 , e2 , ..., ek )) is a function, where each ei ∈ {+, −}. Particularly,
a Smarandachely 2-signed digraph or Smarandachely 2-marked digraph is called abbreviated
a signed digraph or a marked digraph. A signed digraph is an ordered pair S = (D, σ), where
1 Received February 21, 2010. Accepted March 24.
2 The third author is B.E student at Department of Computer Science & Engineering, Rajeev Institute of
Technology, Hassan. This is her first research contribution.
 A Note on Path Signed Digraphs 43

D = (V, A) is a digraph called underlying digraph of S and σ : A → {+, −} is a function. A

marking of S is a function µ : V (D) → {+, −}. A signed digraph S together with a marking µ
is denoted by Sµ . A signed digraph S = (D, σ) is balanced if every semicycle of S is positive
(See [4]). Equivalently, a signed digraph is balanced if every semicycle has an even number of
negative arcs. The following characterization of balanced signed digraphs is obtained in [9].

Proposition 1.1(E. Sampathkumar et al. [9]) A signed digraph S = (D, σ) is balanced if,
and only if, there exist a marking µ of its vertices such that each arc −
→ in S satisfies σ(−
uv → =
uv)
µ(u)µ(v).

In [9], the authors define switching and cycle isomorphism of a signed digraph as follows:

Let S = (D, σ) and S ′ = (D′ , σ ′ ), be two signed digraphs. Then S and S ′ are said to be
isomorphic, if there exists an isomorphism φ : D → D′ (that is a bijection φ : V (D) → V (D′ )
→ is an arc in D then −
such that if −
uv
−−−−−→
φ(u)φ(v) is an arc in D′ ) such that for any arc −→
e ∈ D,
−
→ ′ −
→
σ( e ) = σ (φ( e )).

Given a marking µ of a signed digraph S = (D, σ), switching S with respect to µ is

the operation changing the sign of every arc − → of S ′ by µ(u)σ(−
uv →
uv)µ(v). The signed digraph
obtained in this way is denoted by Sµ (S) and is called µ switched signed digraph or just switched
signed digraph.
Further, a signed digraph S switches to signed digraph S ′ (or that they are switching
equivalent to each other), written as S ∼ S ′ , whenever there exists a marking of S such that
Sµ (S) ∼
= S ′.
Two signed digraphs S = (D, σ) and S ′ = (D′ , σ ′ ) are said to be cycle isomorphic, if there
exists an isomorphism φ : D → D′ such that the sign σ(Z) of every semicycle Z in S equals to
the sign σ(φ(Z)) in S ′ .

Proposition 1.2(E. Sampathkumar et al. [9]) Two signed digraphs S1 and S2 with the same
underlying graph are switching equivalent if, and only if, they are cycle isomorphic.

§2. Path Signed Digraphs

In [3], Harary and Norman introduced the notion of line digraphs for digraphs. The line digraph
L(D) of a given digraph D = (V, A) has the arc set A := A(D) of D for its vertex set and (e, f )
is an arc in L(D) whenever the arcs e and f in D have a vertex in common in such a way that
it is the head of e and the tail of f ; hence, a given digraph H is called a line digraph if there
exists a digraph D such that L(D) ∼ = H. By a natural way, Broersma and Li [1] generalized
the concept of line digraphs to that of directed path graphs.
−→ −
→
Let k be a positive integer, and denote Pk or Ck a directed path or a directed cycle on
−→
k vertices, respectively. Let D be a digraph containing at least one directed path Pk . Denote
−
→ −→ −
→
Πk (D), the set of all Pk ’s of D. Then the directed Pk -graph of D, denoted by Pk (D), is
−
→ −−−→
the digraph with vertex set Πk (D); pq is an arc of Pk (D) if, and only if, there is a Pk+1 or
−→ −→
Ck = (v1 v2 ...vk+1 ) in D (with v1 = vk+1 in the case of a Ck ) such that p = v1 v2 ...vk and
44 P.Siva Kota Reddy, S. Vijay and H. C. Savithri

−
→ −
→
q = v2 ...vk vk+1 . Note that P1 (D) = D and L (D). In [7], the authors proposed an open
−
→
problem for further study, i.e., how to give a characterization for directed P3 -graphs.
−
→
We extend the notion of Pk (D) to the realm of signed digraphs. In a signed digraph
S = (D, σ), where D = (V, A) is a digraph called underlying digraph of S and σ : A → {+, −}
−
→ −→
is a function. The path signed digraph Pk (S) = (Pk (D), σ ′ ) of a signed digraph S = (D, σ)
−
→
is a signed digraph whose underlying digraph is Pk (D) called path digraph and sign of any
→−
− → −
→ →−
− → −→ − →
arc e = Pk Pk′ in Pk (S) is σ ′ (Pk Pk′ ) = σ(Pk )σ(Pk′ ). Further, a signed digraph S = (G, σ)
−
→
is called path signed digraph, if S ∼= Pk (S ′ ), for some signed digraph S ′ . At the end of this
−
→
section, we discuss the structural characterization of path signed digraphs Pk (S).We now gives
a straightforward, yet interesting, property of path signed digraphs.
−
→
Proposition 2.1 For any signed digraph S = (D, σ), its path signed digraph Pk (S) is balanced.
→−
− → −
→ → −
− →
Proof Since sign of any arc σ ′ (e = Pk Pk′ ) in Pk (S) is σ(Pk )σ(Pk′ ), where σ is the marking
−
→ −
→
of Pk (S), by Proposition 1.1, Pk (S) is balanced. 

Remark: For any two signed digraphs S and S ′ with same underlying digraph, their path
signed digraphs are switching equivalent.

In [9], the authors defined line signed digraph of a signed digraph S = (D, σ) as follows:

A line signed digraph L(S) of a signed digraph S = (D, σ) is a signed digraph L(S) =
−
→ −
→ −
→
(L(D), σ ′ ) where for any arc ee′ in L(D), σ ′ (ee′ ) = σ(−
→
e )σ( e′ ) (see also, E. Sampathkumar et
al. [8]).

Hence, we shall call a given signed digraph S a line signed digraph if it is isomorphic to the
line signed digraph L(S ′ ) of some signed digraph S ′ . By the definition of path signed digraphs,
−
→
we observe that P2 (S) = L(S).
−
→
Corollary 2.2 For any signed digraph S = (G, σ), its P2 (S) (=L(S)) is balanced.

In [9], the authors obtain structural characterization of line signed digraphs as follows:

Proposition 2.3(E. Sampathkumar et al. [9]) A signed digraph S = (D, σ) is a line signed
−
→
digraph (or P2 -signed digraph) if, and only if, S is balanced signed digraph and its underlying
−→
digraph D is a line digraph (or P2 -digraph).

Proof Suppose that S is balanced and D is a line digraph. Then there exists a digraph
D′ such that L(D′ ) ∼
= D. Since S is balanced, by Proposition 1.1, there exists a marking µ
of D such that each arc −→ in S satisfies σ(−
uv → = µ(u)µ(v). Now consider the signed digraph
uv)
S = (D , σ ), where for any arc e in D , σ (→
′ ′ ′ −
→ ′ ′ −
e ) is the marking of the corresponding vertex in
′ ∼
D. Then clearly, L(S ) = S. Hence S is a line signed digraph.
Conversely, suppose that S = (D, σ) is a line signed digraph. Then there exists a signed
digraph S ′ = (D′ , σ ′ ) such that L(S ′ ) ∼
= S. Hence D is the line digraph of D′ and by Corollary
2.2, S is balanced. 
We strongly believe that the above Proposition can be generalized to path signed digraphs
 A Note on Path Signed Digraphs 45

−
→
Pk (S) for k ≥ 3. Hence, we pose it as a problem:

Problem 2.4 If S = (D, σ) is a balanced signed digraph and its underlying digraph D is a path
digraph, then S is a path signed digraph.

§3. Switching Equivalence of Signed Digraphs and Path Signed Digraphs

−
→
Broersma and Li [1] concluded that the only connected digraphs D with P3 (D) ∼ = D consists
of a directed cycle with in-trees or out-trees attached to its vertices, with at most non-trivial
trees, where a directed tree T of D is an out-tree of D if V (T ) = V (D) and precisely one vertex
of T has in-degree zero (the root of T ), while all other vertices of T have in-degree one, and an
in-tree of D is defined analogously with respect to out-degrees.

Proposition 3.1(Broersma and Hoede [1]) Let D be connected digraph without sources or
−
→ −
→
sinks. If D has an in-tree or out-tree, then P3 (D) ∼
= D if, and only if, D ∼
= Cn for some n ≥ 3.
−
→ −→
Hence, if D is strongly connected, then P3 (D) ∼ = D if, and only if, D ∼
= Cn for some n ≥ 3.

In the view of the above result, we now characterize signed digraphs that are switching
−
→
equivalent to their P3 -signed digraphs.
−
→
Proposition 3.2 For any strongly connected signed digraph S = (D, σ), S ∼ P3 (S) if, and
−
→
only if, S is balanced and D ∼
= Cn for some n ≥ 3.vskip 3mm
∼ L(D) and hence by Proposition 3.1, D ∼ −
→
Proof Suppose S ∼ L(S). This implies, D = = Cn .
Now, if S is signed digraph, then by Corollary 2.2, implies that L(S) is balanced and hence if
S is unbalanced its line signed digraph L(S) being balanced cannot be switching equivalent to
S in accordance with Proposition 1.2. Therefore, S must be balanced.
−
→ −
→
Suppose that S is balanced and D ∼ = Cn for some n ≥ 3. Then, by Proposition 2.1, P3 (S)
is balanced, the result follows from Proposition 1.2. 
n+k n
In [9], the authors defined a signed digraph S is periodic, if L (S) ∼ L (S) for some
positive integers n and k.
−→
Analogous to the line signed digraphs, we defined periodic for P3 (S) as follows:
−
→
For some positive integers n and k, define that a path signed digraph P3 (S) is periodic, if
−−n+k
−→ −→
P3 (S) ∼ P3n (S).
−→
Proposition 3.3(Broersma and Hoede [1]) If D is strongly connected digraph and P3n (D) ∼
=D
−
→ ∼
for some n ≥ 1, then P3 (D) = D and D is a directed cycle.

The following result is follows from Propositions 2.1,3.2 and 3.3.

−→
Proposition 3.4 If S is strongly connected signed digraph, and P3n (S) ∼ S for some n ≥ 1,
−
→
then P3 (S) ∼ S and D is a directed cycle.

The negation η(S) of a given signed digraph S defined as follows: η(S) has the same
underlying digraph as that of S with the sign of each arc opposite to that given to it in S.
46 P.Siva Kota Reddy, S. Vijay and H. C. Savithri

However, this definition does not say anything about what to do with nonadjacent pairs of
vertices in S while applying the unary operator η(.) of taking the negation of S.
−
→
For a signed digraph S = (D, σ), the Pk (S) is balanced (Proposition 2.1). We now examine,
−
→ −→
the condition under which negation of Pk (S) (i.e., η(Pk (S))) is balanced.
−
→ −
→
Proposition 3.5 Let S = (D, σ) be a signed digraph. If Pk (D) is bipartite then η(Pk (S)) is
balanced.
−
→ −
→
proof Since, by Proposition 2.1, Pk (S) is balanced, then every semicycle in Pk (S) contains
−→
even number of negative arcs. Also, since Pk (G) is bipartite, all semicycles have even length;
−→
thus, the number of positive arcs on any semicycle C in Pk (S) are also even. This implies that
−
→ −
→
the same thing is true in negation of Pk (S). Hence η(Pk (S)) is balanced. 
Proposition 3.2 provides easy solutions to three other signed digraph switching equivalence
relations, which are given in the following results.
−
→
Corollary 3.6 For any signed digraph S = (D, σ), η(S) ∼ P3 (S) if, and only if, S is an
unbalanced signed digraph on any odd semicycle.
−
→ −
→
Corollary 3.7 For any signed digraph S = (D, σ) and for any integer k ≥ 1, Pk (η(S)) ∼ Pk (S).

References

[1] H. J. Broersma and X. Li, Isomorphisms and traversability of directed path graphs, Discuss.
Math. Graph Theory, 22 (2002), 215-228.
[2] J. A. Bondy and U. S. R. Murty, Graph Theory with Applications, MacMillan, London,
1976.
[3] F. Harary and R. Z. Norman, Some properties of line digraphs, Rend. Cric. Mat. Palermo,
2 (9) (1960), 161-168.
[4] F.Harary, R. Norman and D. Cartwright, Structural models: An introduction to the theory
of directed graphs, J. Wiley, New York, 1965.
[5] H. Li and Y. Lin, On the characterization of path graphs, J. Graph Theory, 17 (1993),
463-466.
[6] X. Li and B. Zhao, Isomorphisms of P4 -graphs, Australas. J. Combin., 15 (1997), 135-143.
[7] X. Li and Y. Liu, Path graphs versus line graphs-A survey-Preprint.
[8] E. Sampathkumar, P. Siva Kota Reddy, and M. S. Subramanya, The Line n-sigraph of a
symmetric n-sigraph, Southeast Asian Bull. Math.(Springer-Verlag), to appear.
[9] E. Sampathkumar, M. S. Subramanya and P. Siva Kota Reddy, Characterization of Line
Sidigraphs, Southeast Asian Bull. Math.(Springer-Verlag), to appear.
[10] T. Zaslavsky, Signed Graphs, Discrete Appl. Math., 4(1)(1982), 47-74.
Math.Combin.Book Ser. Vol.1 (2010), 47-64

A Combinatorial Decomposition of
Euclidean Spaces Rn with Contribution to Visibility

Linfan MAO

(Chinese Academy of Mathematics and System Science, Beijing 100080, P.R.China)

E-mail: [email protected]

Abstract: The visibility of human beings only allows them to find objects in R3 at a time
t. That is why physicists prefer to adopt the Euclidean space R3 being physical space of
particles until last century. Recent progress shows the geometrical space of physics maybe
Rn for n ≥ 4, for example, n = 10, or 11 in string theory. Then how to we visualize an object
in Rn for n ≥ 4? This paper presents a combinatorial model, i.e., combinatorial Euclidean
spaces established on Euclidean spaces R3 and prove any such Euclidean space Rn with
n ≥ 4 can be decomposed into such combinatorial structure. We also discuss conditions for
realization Rn in mathematics or physical space by combinatorics and show the space R10
in string theory is a special case in such model.

Key Words: Smarandache multi-space, combinatorial Euclidean space, combinatorial fan-

space, spacetime, p-brane, parallel probe, ultimate theory for the Universe.

AMS(2000): 05A18, 05B40, 83C05

§1. Introduction

A Euclidean space Rn is the point set {(x1 , x2 , · · · , xn )| xi ∈ R, 1 ≤ i ≤ n} for an integer

n ≥ 1. The structure of our eyes determines that one can only detect particles in an Euclidean
space R3 , which gave rise to physicists prefer R3 as a physical space. In fact, as showed in the
references [2], [18] and [21], our visible geometry is the spherical geometry. This means that
we can only observe parts of a phenomenon in the Universe if its topological dimensional≥ 4
([1], [14]). It should be noted that if parallel worlds [6], [20] exist the dimensional of Universe
must≥ 4. Then,

Can we establish a model for detecting behaviors of particles in Rn with n ≥ 4?

This paper suggests a combinatorial model and a system for visualizing phenomenons in the
space Rn with n ≥ 4. For this object, we establish the decomposition of Rn underlying a
connected graph G in Sections 2 and 3, then show how to establish visualizing system in such
combinatorial model and acquire its global properties, for example, the Einstein’s gravitational
equations in Section 4. The final sections discusses conditions of its physical realization. Termi-
nologies and notations not defined here are followed in [1], [3] and [4] for topology, gravitational
fields and graphs.
1 Received January 16, 2010. Accepted March 24, 2010.
48 L.F.Mao

§2. Combinatorial Euclidean Spaces

Definition 2.1([13]) A combinatorial system CG is a union of mathematical systems (Σ1 ; R1 ),(Σ2 ; R2 ),

· · · , (Σm ; Rm ) for an integer m, i.e.,

m
[ m
[
CG = ( Σi ; Ri )
i=1 i=1

with an underlying connected graph structure G, i.e., a particular Smarandache multi-space([8]),

where

V (G) = {Σ1 , Σ2 , · · · , Σm },
T
E(G) = { (Σi , Σj ) | Σi Σj 6= ∅, 1 ≤ i, j ≤ m}.

Definition 2.2 A combinatorial Euclidean space is a combinatorial system CG of Euclidean

spaces Rn1 , Rn2 , · · · , Rnm with an underlying structure G, denoted by EG (n1 , · · · , nm ) and
abbreviated to EG (r) if n1 = · · · = nm = r.

It should be noted that a combinatorial Euclidean space is itself a Euclidean space. This
fact enables us to decomposition a Euclidean space Rn into Euclidean spaces Rn1 , Rn2 , · · · ,
Rnm underlying a graph G but with less dimensions, which gives rise to a packing problem on
Euclidean spaces following.

Problem 2.1 Let Rn1 , Rn2 , · · · , Rnm be Euclidean spaces. In what conditions do they consist
of a combinatorial Euclidean space EG (n1 , · · · , nm )?

Notice that a Euclidean space Rn is an n-dimensional vector space with a normal basis
ǫ1 = (1, 0, · · · , 0), ǫ2 = (0, 1, 0 · · · , 0), · · · , ǫn = (0, · · · , 0, 1), namely, it has n orthogonal
orientations. So if we think any Euclidean space Rn is a subspace of a Euclidean space Rn∞
with a finite but sufficiently large dimension n∞ , then two Euclidean spaces Rnu and Rnv have
a non-empty intersection if and only if they have common orientations. Whence, we only need
to determine the number of different orthogonal orientations in EG (n1 , · · · , nm ).
Denoted by Xv1 , Xv2 , · · · , Xvm consist of these orthogonal orientations in Rnv1 , Rnv2 ,
· · · , Rnvm , respectively. An intersection graph G[Xv1 , Xv2 , · · · , Xvm ] of Xv1 , Xv2 , · · · , Xvm is
defined by ([5])

V (G[Xv1 , Xv2 , · · · , Xvm ]) = {v1 , v2 , · · · , vm },

T
E[Xv1 , Xv2 , · · · , Xvm ] = {(vi , vj )|Xvi Xvj 6= ∅, 1 ≤ i 6= j ≤ m}.

By definition, we know that

G ∼
= G[Xv1 , Xv2 , · · · , Xvm ],

which transfers the Problem 2.1 of Euclidean spaces to a combinatorial one following.

Problem 2.2 For given integers κ, m ≥ 2 and n1 , n2 , · · · , nm , find finite sets Y1 , Y2 , · · · , Ym

with their intersection graph being G such that |Yi | = ni , 1 ≤ i ≤ m, and |Y1 ∪Y2 ∪· · · ∪Ym | = κ.
 A Combinatorial Decomposition of Euclidean Spaces Rn with Contribution to Visibility 49

2.1 The maximum dimension of EG (n1 , · · · , nm )

First, applying the inclusion-exclusion principle, we get the next counting result.

Theorem 2.1 Let EG (n1 , · · · , nm ) be a combinatorial Euclidean space of Rn1 , Rn2 , · · · , Rnm
with an underlying structure G. Then
P T T T
dimEG (n1 , · · · , nm ) = (−1)s+1 dim(Rnv1 Rnv2 ··· Rnvs ),
hvi ∈V (G)|1≤i≤si∈CLs (G)

where nvi denotes the dimensional number of the Euclidean space in vi ∈ V (G) and CLs (G)
consists of all complete graphs of order s in G.

Proof By definition, Rnu ∩ Rnv 6= ∅ only if there is an edge (Rnu , Rnv ) in G. This
condition can be generalized to a more general situation, i.e., Rnv1 ∩ Rnv2 ∩ · · · ∩ Rnvl 6= ∅ only
if hv1 , v2 , · · · , vl iG ∼
= Kl .
In fact, if Rnv1 ∩ Rnv2 ∩ · · · ∩ Rnvl 6= ∅, then Rnvi ∩ Rnvj 6= ∅, which implies that
(Rnvi , Rnvj ) ∈ E(G) for any integers i, j, 1 ≤ i, j ≤ l. Therefore, hv1 , v2 , · · · , vl iG is a complete
graph of order l in the intersection graph G.
Now we are needed to count these orthogonal orientations in EG (n1 , · · · , nm ). In fact, the
number of different orthogonal orientations is
S
dimEG (n1 , · · · , nm ) = dim( Rnv )
v∈V (G)

by previous discussion. Applying the inclusion-exclusion principle, we find that

[
dimEG (n1 , · · · , nm ) = dim( Rnv )
v∈V (G)
X \ \ \
= (−1)s+1 dim(Rnv1 Rnv2 ··· Rnvs )
{v1 ,··· ,vs }⊂V (G)
X \ \ \
= (−1)s+1 dim(Rnv1 Rnv2 ··· Rnvs ).
hvi ∈V (G)|1≤i≤si∈CLs (G)


nv1 nv2 nvs nv1 nv2
Notice that dim(R ∩R ∩ · · · ∩ R ) = nv1 if s = 1 and dim(R ∩ R ) 6= 0 only
if (Rnv1 , Rnv2 ) ∈ E(G). We get a more applicable formula for calculating dimEG (n1 , · · · , nm )
on K3 -free graphs G by Theorem 2.1.

Corollary 2.1 If G is K3 -free, then

P P T
dimEG (n1 , · · · , nm ) = nv − dim(Rnu Rnv ).
v∈V (G) (u,v)∈E(G)

Particularly, if G = v1 v2 · · · vm a circuit for an integer m ≥ 4, then

m
P m
P T
dimEG (n1 , · · · , nm ) = nvi − dim(Rnvi Rnvi+1 ),
i=1 i=1

where each index is modulo m.

50 L.F.Mao

Now we determine the maximum dimension of combinatorial Euclidean spaces of Rn1 ,

R , · · · , Rnm with an underlying structure G.
n2

Theorem 2.2 Let EG (nv1 , · · · , nvm ) be a combinatorial Euclidean space of Rnv1 , Rnv2 , · · · ,
Rnvm with an underlying graph G, V (G) = {v1 , v2 , · · · , vm }. Then the maximum dimension
dimmax EG (nv1 , · · · , nvm ) of EG (nv1 , · · · , nvm ) is
P
dimmax EG (nv1 , · · · , nvm ) = 1 − m + nv
v∈V (G)

with conditions dim(Rnu ∩ Rnv ) = 1 for ∀(u, v) ∈ E(G).

Proof Let Xv1 , Xv2 , · · · , Xvm consist of these orthogonal orientations in Rnv1 , Rnv2 , · · · ,
Rnvm , respectively. Notice that
[ \
|Xvi Xvj | = |Xvi | + |Xvj | − |Xvi Xvj |
for 1 ≤ i 6= j ≤ m by Theorem 1.5.1 in the case of n = 2. We immediately know that |Xvi ∪Xvj |
attains its maximum value only if |Xvi ∩ Xvj | is minimum. Since Xvi and Xvj are nonempty
sets, we find that the minimum value of |Xvi ∩ Xvj | = 1 if (vi , vj ) ∈ E(G).
The proof is finished by the inductive principle. Not loss of generality, assume (v1 , v2 ) ∈
S
E(G). Then we have known that |Xv1 Xv2 | attains its maximum

|Xv1 | + |Xv2 | − 1
only if |Xv1 ∩ Xv2 | = 1. Since G is connected, not loss of generality, let v3 be adjacent with
{v1 , v2 } in G. Then by
[ [ [ [ \
|Xv1 Xv2 Xv3 | = |Xv1 Xv2 | + |Xv3 | − |(Xv1 Xv2 ) Xv3 |,
we know that |Xv1 ∪ Xv2 ∪ Xv3 | attains its maximum value only if |Xv1 ∪ Xv2 | attains its
maximum and |(Xv1 ∪ Xv2 ) ∩ Xv3 | = 1 for (Xv1 ∪ Xv2 ) ∩ Xv3 6= ∅. Whence, |Xv1 ∩ Xv3 | = 1 or
|Xv2 ∩ Xv3 | = 1, or both. In the later case, there must be |Xv1 ∩ Xv2 ∩ Xv3 | = 1. Therefore,
the maximum value of |Xv1 ∪ Xv2 ∪ Xv3 | is

|Xv1 | + |Xv2 | + |Xv3 | − 2.

Generally, we assume the maximum value of |Xv1 ∪ Xv2 ∪ · · · ∪ Xvk | to be

|Xv1 | + |Xv2 | + · · · + |Xvk | − k + 1

for an integer k ≤ m with conditions |Xvi ∩ Xvj | = 1 hold if (vi , vj ) ∈ E(G) for 1 ≤ i 6= j ≤ k.
By the connectedness of G, without loss of generality, we choose a vertex vk+1 adjacent with
{v1 , v2 , · · · , vk } in G and find out the maximum value of |Xv1 ∪ Xv2 ∪ · · · ∪ Xvk ∪ Xvk+1 |. In
fact, since

|Xv1 ∪ Xv2 ∪ · · · ∪ Xvk ∪ Xvk+1 | = |Xv1 ∪ Xv2 ∪ · · · ∪ Xvk | + |Xvk+1 |

\
− |(Xv1 ∪ Xv2 ∪ · · · ∪ Xvk ) Xvk+1 |,
 A Combinatorial Decomposition of Euclidean Spaces Rn with Contribution to Visibility 51

we know that |Xv1 ∪ Xv2 ∪ · · · ∪ Xvk ∪ Xvk+1 | attains its maximum value only if |Xv1 ∪ Xv2 ∪
T
· · · ∪ Xvk | attains its maximum and |(Xv1 ∪ Xv2 ∪ · · · ∪ Xvk ) Xvk+1 | = 1 for (Xv1 ∪ Xv2 ∪ · · · ∪
Xvk ) ∩ Xvk+1 6= ∅. Whence, |Xvi ∩ Xvk+1 | = 1 if (vi , vk+1 ) ∈ E(G). Consequently, we find that
the maximum value of |Xv1 ∪ Xv2 ∪ · · · ∪ Xvk ∪ Xvk+1 | is

|Xv1 | + |Xv2 | + · · · + |Xvk | + |Xvk+1 | − k.

Notice that our process searching for the maximum value of |Xv1 ∪ Xv2 ∪ · · · ∪ Xvk | does
not alter the intersection graph G of Xv1 , Xv2 , · · · , Xvm . Whence, by the inductive principle
we finally get the maximum dimension dimmax EG of EG , that is,

dimmax EG (nv1 , · · · , nvm ) = 1 − m + n1 + n2 + · · · + nm

with conditions dim(Rnu ∩ Rnv ) = 1 for ∀(u, v) ∈ E(G). 

2.2 The minimum dimension of EG (n1 , · · · , nm )

Determining the minimum value dimmin EG (n1 , · · · , nm ) of EG (n1 , · · · , nm ) is a difficult problem

in general case. But for some graph families we can determine its minimum value.

Theorem 2.3 Let EG (nv1 , nv2 , · · · , nvm ) be a combinatorial Euclidean space of Rnv1 , Rnv2 , · · · ,
Rnvm with an underlying graph G, V (G) = {v1 , v2 , · · · , vm } and {v1 , v2 , · · · , vl } an independent
vertex set in G. Then

l
X
dimmin EG (nv1 , · · · , nvm ) ≥ nvi
i=1

and with the equality hold if G is a complete bipartite graph K(V1 , V2 ) with partite sets V1 =
{v1 , v2 , · · · , vl }, V2 = {vl+1 , vl+2 , · · · , vm } and

l
X m
X
nvi ≥ nvi .
i=1 i=l+1

Proof Similarly, we use Xv1 , Xv2 , · · · , Xvm to denote these orthogonal orientations in Rnv1 ,
nv2
R , · · · , Rnvm , respectively. By definition, we know that
T
Xvi Xvj = ∅, 1 ≤ i 6= j ≤ l
for (vi , vj ) 6∈ E(G). Whence, we get that

m
[ l
[ l
X
| Xvi | ≥ | Xvi | = nvi .
i=1 i=1 i=1

By the assumption,

l
X m
X
nvi ≥ nvi ,
i=1 i=l+1

we can partition Xv1 , Xv2 , · · · , Xvm to

52 L.F.Mao

S
m S
Xv1 = ( Yi (v1 )) Z(v1 ),
i=l+1

S
m S
Xv2 = ( Yi (v2 )) Z(v2 ),
i=l+1
···············,
m
S S
Xvl =( Yi (vl )) Z(vl )
i=l+1

l
P
such that |Yi (vk )| = |Xvi | for any integer i, l + 1 ≤ i ≤ m, where Z(vi ) maybe an empty
k=1
set for integers i, 1 ≤ i ≤ l. Whence, we can choose
l
S
Xv′ i = Yi (vk )
k=1

to replace each Xvi for any integer i, 1 ≤ i ≤ m. Notice that the intersection graph of
Xv1 , Xv2 , · · · , Xvl , Xv′ l+1 , · · · , Xv′ m is still the complete bipartite graph K(V1 , V2 ), but
m
S l
S l
P
| Xvi | = | Xvi | = ni .
i=1 i=1 i=1

Therefore, we get that

P
l
dimmin EG (nv1 , · · · , nvm ) = nvi
i=1

in the case of complete bipartite graph K(V1 , V2 ) with partite sets V1 = {v1 , v2 , · · · , vl }, V2 =
{vl+1 , vl+2 , · · · , vm } and
l
P m
P
nvi ≥ nvi . 
i=1 i=l+1

Although the lower bound of dimEG (nv1 , · · · , nvm ) in Theorem 2.3 is sharp, but it is not
better if G is given in some cases. Consider a complete system of r-subsets of a set with less
than 2r elements. We know the next conclusion if G = Km .

Theorem 2.4 For any integer r ≥ 2, let EKm (r) be a combinatorial Euclidean space of
Rr , · · · , Rr , and there exists an integer s, 0 ≤ s ≤ r − 1 such that
| {z }
m
   
r+s−1 r+s
 <m≤ .
r r

Then

dimmin EKm (r) = r + s.

Proof We denote by X1 , X2 , · · · , Xm these sets consist of orthogonal orientations in m

Euclidean spaces Rr . Then each Xi , 1 ≤ i ≤ m, is an r-set. By assumption,
 A Combinatorial Decomposition of Euclidean Spaces Rn with Contribution to Visibility 53

   
r+s−1 r+s
 <m≤ 
r r
and 0 ≤ s ≤ r − 1, we know that two r-subsets of an (r + s)-set must have a nonempty
intersection. So we can determine these m r-subsets X1 , X2 , · · · , Xm by using the complete
system of r-subsets in an (r + s)-set, and these m r-subsets X1 , X2 , · · · , Xm can not be chosen
[m
in an (r + s − 1)-set. Therefore, we find that | Xi | = r + s, i.e., if 0 ≤ s ≤ r − 1, then
i=1
dimmin EKm (r) = r + s. 

For general combinatorial spaces EKm (nv1 , · · · , nvm ) of Rnv1 , Rnv2 , · · · , Rnvm , we get
their minimum dimension if nvm is large enough.

Theorem 2.5 Let EKm be a combinatorial Euclidean space of Rnv1 , Rnv2 , · · · , Rnvm , nv1 ≥
nv2 ≥ · · · ≥ nvm ≥ ⌈log2 ( 2nv1m+1
−nv2
−1
)⌉ + 1 and V (Km ) = {v1 , v2 , · · · , vm }. Then

m+1
dimmin EKm (nv1 , · · · , nvm ) = nv1 + ⌈log2 ( )⌉.
2nv1 −nv2 −1

Proof Let Xv1 , Xv2 , · · · , Xvm be sets consist of these orthogonal orientations in Rnv1 ,
Rnv2 , · · · , Rnvm , respectively and

m
2s−1 < + 1 ≤ 2s
2k+1
−1
for an integer s, where k = nv1 − nv2 . Then we find that

m+1
⌈log2 ( nv1 −nv2 −1 )⌉
= s.
2
We construct a family {Yv1 , Yv2 , · · · , Yvm } with none being a subset of another, |Yvi | = |Xvi |
for 1 ≤ i ≤ m and its intersection graph is still Km , but with
[ [ [
|Yv1 Yv2 ··· Yvm | = nv1 + s.
In fact, let Xv1 = {x1 , x2 , · · · , xnv2 , xnv2 +1 , · · · , xnv1 } and U = {u1 , u2 , · · · , us }, such as
those shown in Fig.2.1 for s = 1 and nv1 = 9.

u1
X2 X3 X4

x1 x2 x3 x4 x5 x6 x7 x8 x9

X1

Fig.2.1
Choose g elements xi1 , xi2 , · · · , xig ∈ Xv1 and h ≥ 1 elements uj1 , uj2 , · · · , ujh ∈ U . We
construct a finite set
54 L.F.Mao

Xg.h = {xi1 , xi2 , · · · , xig , uj1 , uj2 , · · · , ujh }

with a cardinal g + h. Let g + h = |Xv1 |, |Xv2 |, · · · , |Xvm |, respectively. We consequently

find such sets Yv1 , Yv2 , · · · , Yvm . Notice that there are no one set being a subset of another
in the family {Yv1 , Yv2 , · · · , Yvm }. So there must have two elements in each Yvi , 1 ≤ i ≤ m
at least such that one is in U and another in {xnv2 , xnv2 +1 , · · · , xnv1 }. Now since nvm ≥
⌈log2 ( 2nv1m+1
−nv2
−1
)⌉ + 1, there are
  
k+1
XX s
k+1 s
   = (2k+1 − 1)(2s − 1) ≥ m
i=1 j=1 i j

different sets Yv1 , Yv2 , · · · , Yvm altogether with |Xv1 | = |Yv1 |, · · · , |Xvm | = |Yvm |. None of them
is a subset of another and their intersection graph is still Km . For example,

Xv1 , {u1 , x1 , · · · , xnv2 −1 },

{u1 , xnv2 −nv3 +2 , · · · , xnv2 },

··················,

{u1 , xnvk−1 −nvk +2 , · · · , xnvk }

are such sets with only one element u1 in U . See also in Fig.4.1.1 for details. It is easily to
know that

[ [ [ m+1
|Yv1 Yv2 ··· Yvm | = nv1 + s = nv1 + ⌈log2 ( )⌉
2nv1 −nv2 − 1
in our construction.
Conversely, if there exists a family {Yv1 , Yv2 , · · · , Yvm } such that |Xv1 | = |Yv1 |, · · · , |Xvm | =
|Yvm | and

[ [ [
|Yv1 Yv2 ··· Yvm | < nv1 + s,

then there at most

  
k+1
XX s
k+1 s−1
   = (2k+1 − 1)(2s−1 − 1) < m
i=1 j=1 i j

different sets in {Yv1 , Yv2 , · · · , Yvm } with none being a subset of another. This implies that
there must exists integers i, j, 1 ≤ i 6= j ≤ m with Yvi ⊂ Yvj , a contradiction. Therefore, we get
the minimum dimension dimmin EKm of EKm to be

m+1
dimmin EKm (nv1 , · · · , nvm ) = nv1 + ⌈log2 ( )⌉.
2nv1 −nv2 − 1


As we introduce in Section 1, the combinatorial space of R3 is particularly interested in

physics. In the case of Km , we can determine its minimum dimension.
 A Combinatorial Decomposition of Euclidean Spaces Rn with Contribution to Visibility 55

Theorem 2.5 Let EKm (3) be a combinatorial Euclidean space of R3 , · · · , R3 . Then

| {z }
m


 3, if m = 1,



 4, if 2 ≤ m ≤ 4,
dimmin EKm (3) =
 5,
 if 5 ≤ m ≤ 10,



 √
2 + ⌈ m⌉, if m ≥ 11.

Proof Let X1 , X2 , · · · , Xm be these sets consist of orthogonal orientations in m Euclidean

spaces R3 , respectively and |X1 ∪ X2 ∪ · · · ∪ Xm | = l. Then each Xi , 1 ≤ i ≤ m, is a 3-set.
 
5
In the case of m ≤ 10 =  , any s-sets have a nonempty intersection. So it is easily
2
to check that



 3, if
 m = 1,
dimmin EKm (3) = 4, if 2 ≤ m ≤ 4,



5, if 5 ≤ m ≤ 10.

We only consider the case of m ≥ 11. Let X = {u, v, w} be a chosen 3-set. Notice that
any 3-set will intersect X with 1 or 2 elements. Our discussion is divided into three cases.

Case 1 There exist 3-sets X1′ , X2′ , X3′ such that X1′ ∩ X = {u, v}, X2′ ∩ X = {u, w} and
X3′ ∩ X = {v, w} such as those shown in Fig.2.2, where each triangle denotes a 3-set.

v
u w

Fig.2.2

Notice that there are no 3-sets X ′ such that |X ′ ∩ X| = 1 in this case. Otherwise, we
can easily find two 3-sets with an empty intersection, a contradiction. Counting such 3-sets,
we know that there are at most 3(v − 3) + 1 3-sets with their intersection graph being Km .
Thereafter, we know that

m−1
m ≤ 3(l − 3) + 1, i.e., l≥⌈ ⌉ + 3.
3

Case 2 There are 3-sets X1′ , X2′ but no 3-set X3′ such that X1′ ∩ X = {u, v}, X2′ ∩ X = {u, w}
and X3′ ∩ X = {v, w} such as those shown in Fig.2.3, where each triangle denotes a 3-set.
56 L.F.Mao

v
u w

Fig.2.3

In this case, there are no 3-sets X ′ such that X ′ ∩ X = {u} or {w}. Otherwise, we can
easily find two 3-sets with an empty intersection, a contradiction. Enumerating such 3-sets, we
know that there are at most
 
l−3
2(l − 1) +  +1
2

3-sets with their intersection graph still being Km . Whence, we get that
 
√
l−3 3+ 8m + 17
m ≤ 2(l − 1) +   + 1, i.e., l≥⌈ ⌉.
2 2

Case 3 There are a 3-set X1′ but no 3-sets X2′ , X3′ such that X1′ ∩ X = {u, v}, X2′ ∩ X = {u, w}
and X3′ ∩ X = {v, w} such as those shown in Fig.2.4, where each triangle denotes a 3-set.

u v
w

Fig.2.4

Enumerating 3-sets in this case, we know that there are at most

 
l−2
l −2+ 2 
2

such 3-sets with their intersection graph still being Km . Therefore, we find that
 
l−2 √
m ≤ l −2+2 , i.e., l ≥ 2 + ⌈ m⌉.
2

Combining these Cases 1 − 3, we know that

√
m−1 3+ 8m + 17 √ √
l ≥ min{⌈ ⌉ + 3, ⌈ ⌉, 2 + ⌈ m⌉} = 2 + ⌈ m⌉.
3 2
 A Combinatorial Decomposition of Euclidean Spaces Rn with Contribution to Visibility 57

Conversely, there 3-sets constructed in Case 3 show that there indeed exist 3-sets X1 , X2 , · · · , Xm
whose intersection graph is Km , where
 
l−2
m = l −2+2 .
2

Therefore, we get that

√
dimmin EKm (3) = 2 + ⌈ m⌉

if m ≥ 11. This completes the proof. 

§3. A Combinatorial Model of Euclidean Spaces Rn with n ≥ 4

e 1 , · · · , nm ) is the combinatorial space EK (n1 , · · · , nm ) of Rn1 ,

A combinatorial fan-space R(n m
n2 nm
R , ···, R such that for any integers i, j, 1 ≤ i 6= j ≤ m,

\ m
\
Rni Rnj = Rnk ,
k=1
m
T
which is in fact a p-brane with p = dim Rnk in string theory ([15]-[17]), seeing Fig.3.1 for
k=1
details.

6

 p-brane -

?
Fig.3.1
e 1 , · · · , nm ) we can present it by an m × nm coordinate matrix [x] following
For ∀p ∈ R(n
xl
with xil = m for 1 ≤ i ≤ m, 1 ≤ l ≤ m, b

 
x11 ··· x1m
b x1(m)+1)
b · · · x1n1 ··· 0
 
 x ··· x2m x2(m+1) · · · x2n2 ··· 0 
 21 b b 
[x] =  .
 ··· ··· ··· ··· ··· ··· 
 
xm1 · · · xmm
b xm(m+1)
b ··· ··· xmnm −1 xmnm

By definition, we know the following result.

58 L.F.Mao

e 1 , · · · , nm ) be a fan-space. Then
Theorem 3.1 Let R(n

e 1 , · · · , nm ) = m P
m
dimR(n b+ (ni − m),
b
i=1

where
m
T
m
b = dim( Rnk ). 
k=1

The inner product h(A), (B)i of (A) and (B) is defined by

X
h(A), (B)i = aij bij .
i,j

Then we know the next result by definition.

Theorem 3.2 Let (A), (B), (C) be m × n matrixes and α a constant. Then
(1) hαA, Bi = α hB, Ai;
(2) hA + B, Ci = hA, Ci + hB, Ci;
(3) hA, Ai ≥ 0 with equality hold if and only if (A) = Om×n .

Theorem 3.3 Let (A), (B) be m × n matrixes. Then

2
h(A), (B)i ≤ h(A), (A)i h(B), (B)i

and with equality hold only if (A) = λ(B), where λ is a real constant.

Proof If (A) = λ(B), then hA, Bi2 = λ2 hB, Bi2 = hA, Ai hB, Bi. Now if there are no
constant λ enabling (A) = λ(B), then (A) − λ(B) 6= Om×n for any real number λ. According
to Theorem 3.2, we know that

h(A) − λ(B), (A) − λ(B)i > 0,

i.e.,

h(A), (A)i − 2λ h(A), (B)i + λ2 h(B), (B)i > 0.

Therefore, we find that

∆ = (−2 h(A), (B)i)2 − 4 h(A), (A)i h(B), (B)i < 0,

namely,
2
h(A), (B)i < h(A), (A)i h(B), (B)i . 

Theorem 3.4 For a given integer sequence n1 , n2 , · · · , nm , m ≥ 1 with 0 < n1 < n2 < · · · < nm ,
e 1 , · · · , nm ); d) is a metric space.
(R(n

e 1 , · · · , nm ); d).
Proof We only need to verify that each condition for a metric space is hold in (R(n
e 1 , · · · , nm ), by definition we know that
For two point p, q ∈ R(n
 A Combinatorial Decomposition of Euclidean Spaces Rn with Contribution to Visibility 59

p
d(p, q) = h[p] − [q], [p] − [q]i ≥ 0
with equality hold if and only if [p] = [q], namely, p = q and
p p
d(p, q) = h[p] − [q], [p] − [q]i = h[q] − [p], [q] − [p]i = d(q, p).
e 1 , · · · , nm ). By Theorem 3.3, we then find that
Now let u ∈ R(n

(d(p, u) + d(u, p))2

p
= h[p] − [u], [p] − [u]i + 2 h[p] − [u], [p] − [u]i h[u] − [q], [u] − [q]i
+ h[u] − [q], [u] − [q]i
≥ h[p] − [u], [p] − [u]i + 2 h[p] − [u], [u] − [q]i + h[u] − [q], [u] − [q]i
= h[p] − [q], [p] − [q]i = d2 (p, q).

e 1 , · · · , nm ); d) is a metric space.
Whence, d(p, u) + d(u, p) ≥ d(p, q) and (R(n 
e
According to Theorem 3.1, a combinatorial fan-space R(n1 , n2 , · · · , nm ) can be turned
Pm
into a Euclidean space Rn with n = m b + (ni − m). b Now the inverse question is that for
i=1
a Euclidean space Rn , weather there exist a combinatorial Euclidean space EG (n1 , · · · , nm ) of
Euclidean spaces Rn1 , Rn2 , · · · , Rnm such that dimRn1 ∪ Rn2 ∪ · · · ∪ Rnm = n? We get the
following decomposition result of Euclidean spaces.

Theorem 3.5 Let Rn be a Euclidean space, n1 , n2 , · · · , nm integers with m

b < ni < n for
1 ≤ i ≤ m and the equation
m
X
mb+ (ni − m)
b =n
i=1

b 1≤m
hold for an integer m, e 1 , n2 , · · · , nm )
b ≤ n. Then there is a combinatorial fan-space R(n
such that

Rn ∼ e 1 , n2 , · · · , nm ).
= R(n

Proof Not loss of generality, assume the normal basis of Rn is ǫ1 = (1, 0, · · · , 0), ǫ2 =
(0, 1, 0 · · · , 0), · · · , ǫn = (0, · · · , 0, 1). Then its coordinate system of Rn is (x1 , x2 , · · · , xn ).
Since
P
m
n−m
b = (ni − m),
b
i=1

choose

R1 = hǫ1 , ǫ2 , · · · , ǫm
b , ǫm+1
b , · · · , ǫn1 i ;
b , ǫn1 +1 , ǫn1 +2 , · · · , ǫn2 i ;
R2 = hǫ1 , ǫ2 , · · · , ǫm
b , ǫn2 +1 , ǫn2 +2 , · · · , ǫn3 i ;
R3 = hǫ1 , ǫ2 , · · · , ǫm
··························· ;
60 L.F.Mao

Rm = ǫ1 , ǫ2 , · · · , ǫm
b , ǫnm−1 +1 , ǫnm−1 +2 , · · · , ǫnm .

T
m
e 1 , n2 , · · · , nm ) is
Calculation shows that dimRi = ni and dim( Ri ) = m.
b Whence R(n
i=1
a combinatorial fan-space. Whence,

Rn ∼ e 1 , n2 , · · · , nm ).
= R(n 
e 1 , n2 , · · · , nm ) is in fact EK (n1 , n2 , · · · , nm ).
Notice that a combinatorial fan-space R(n m

Let ni = 3 for 1 ≤ i ≤ m. We get a result following by Theorem 3.5.

Corollary3.1 Let Rn be a Euclidean space with n ≥ 4. Then there is a combinatorial Euclidean

space EKm (3) such that

Rn ∼
= EKm (3)
n−1
with m = 2 or m = n − 2.

§4. A Particle in Euclidean Spaces Rn with n ≥ 4

Corollary 3.1 asserts that an Euclidean space Rn can be really decomposed into 3-dimensional
Euclidean spaces R3 underlying a complete graph Km with m = n−1 2 or m = n−2. This suggests
n
that we can visualize a particle in Euclidean space R by detecting its partially behavior in
each R3 . That is to say, we are needed to establish a parallel probe for Euclidean space Rn if
n ≥ 4.
Generally, a parallel probe on a combinatorial Euclidean space EG (n1 , n2 , · · · , nm ) is the
set of probes established on each Euclidean space Rni for integers 1 ≤ i ≤ m, particularly for
EG (3) which one can detects a particle in its each space R3 such as those shown in Fig.4.1 in
where G = K4 and there are four probes P1 , P2 , P3 , P4 .

P1
? ? P2

- R3 R3 

- R3 R3 
P3
6 6 P4

Fig.4.1
 A Combinatorial Decomposition of Euclidean Spaces Rn with Contribution to Visibility 61

Notice that data obtained by such parallel probe is a set of local data F (xi1 , xi2 , xi3 ) for
1 ≤ i ≤ m underlying G, i.e., the detecting data in a spatial ǫ should be same if ǫ ∈ R3u ∩ R3v ,
where R3u denotes the R3 at u ∈ V (G) and (R3u , R3v ) ∈ E(G).
For data not in the R3 we lived, it is reasonable that we can conclude that all are the same
as we obtained. Then we can analyze the global behavior of a particle in Euclidean space Rn
with n ≥ 4.
Then how to apply this speculation? Let us consider the gravitational field with dimensional≥
4. We know the Einstein’s gravitation field equations in R3 are

1
Rµν − gµν R = κTµν ,
2
α
where Rµν = Rµαν = g αβ Rαµβν , R = g µν Rµν are the respective Ricci tensor, Ricci scalar
curvature and

8πG
κ= = 2.08 × 10−48 cm−1 · g −1 · s2
c4
Now for a gravitational field Rn with n ≥ 4, we decompose it into dimensional 3 Euclidean
spaces R3u , R3v , · · · , R3w . Then we find Einstein’s gravitational equations shown in [4] as follows:

1
Rµu νu − gµu νu R = −8πGEµu νu ,
2

1
Rµv νv − gµv νv R = −8πGEµv νv ,
2

··············· ,

1
Rµw νw − gµw νw R = −8πGEµw νw
2
for each R3u , R3v , · · · , R3w . If we decompose Rn into a combinatorial Euclidean fan-space
e 3, · · · , 3), then u, v, · · · , w can be abbreviated to 1, 2 · · · , m. In this case, these gravita-
R(3,
| {z }
m
tional equations can be represented by

1
R(µν)(στ ) − g(µν)(στ ) R = −8πGE(µν)(στ )
2
with a coordinate matrix
 
x11 ··· x1m
b
··· x13
 
 x21 ··· x2m
b
··· x23 
 
[xp ] =  
 ··· ··· ··· ··· ··· 
 
xm1 · · · xmm
b
· · · xm3
m
T m
T xl
for a point p ∈ Rn , where m
b = dim( Rni ) a constant for ∀p ∈ Rni and xil = m for
i=1 i=1
1 ≤ i ≤ m, 1 ≤ l ≤ m.b Because the local behavior is that of the projection of the global.
Whence, the following principle for determining behavior of particles in Rn , n ≥ 4 hold.
62 L.F.Mao

Projective Principle A physics law in a Euclidean space Rn ∼ e 3, · · · , 3) with n ≥ 4 is

= R(3,
| {z }
m
e 3, · · · , 3).
invariant under a projection on R3 in R(3,
| {z }
m

Applying this principe enables us to find a spherically symmetric solution of Einstein’s

gravitational equations in Euclidean space Rn .

§5. Discussions

e 3, · · · , 3)
A simple calculation shows that the dimension of the combinatorial Euclidean fan-space R(3,
| {z }
m
in Section 3 is

e 3, · · · , 3) = 3m + (1 − m)m,
dimR(3, b (4 − 1)
| {z }
m

e 3, · · · , 3) = 6, 9, 12 if m
for example, dimR(3, b = 0 and 5, 7, 9 if m
b = 1 and m = 2, 3, 4 with
| {z }
m
an additional time dimension t.
We have discussed in Section 1 that the visible geometry is the spherical geometry of
dimensional 3. That is why the sky looks like a spherical surface. In these geometrical elements,
such as those of point, line, ray, block, body,· · · , etc., we can only see the image of bodies on
our spherical surface, i.e., surface blocks.
Then what is the geometry of transferring information? Here, the term information in-
cludes information known or not known by human beings. So the geometry of transferring
information consists of all possible transferring routes. In other words, a combinatorial geom-
etry of dimensional≥ 1. Therefore, not all information transferring can be seen by our eyes.
But some of them can be felt by our six organs with the helps of apparatus if needed. For
example, the magnetism or electromagnetism can be only detected by apparatus. Consider m b
the discussion is divided into two cases, which lead to two opposite conclusions following.

Case 1. m
b = 3.
e 3, · · · , 3) = 3, i.e., all Euclidean
In this case, by the formula (4 − 1) we get that dimR(3,
| {z }
m
spaces R31 , R32 , · · · , R3m are in one R3 , which is the most enjoyed case by human beings. If
it is so, all the behavior of Universe can be realized finally by human beings, particularly, the
observed interval is ds and all natural things can be come true by experiments. This also means
that the discover of science will be ended, i.e., we can find an ultimate theory for the Universe
- the Theory of Everything. This is the earnest wish of Einstein himself beginning, and then
more physicists devoted all their lifetime to do so in last century.

Case 2. b ≤ 2.
m
 A Combinatorial Decomposition of Euclidean Spaces Rn with Contribution to Visibility 63

e 3, · · · , 3) ≥ 4. In this case, the observed interval in the

If the Universe is so, then dimR(3,
| {z }
m
field R3human where human beings live is

ds2human = a(t, r, θ, φ)dt2 − b(t, r, θ, φ)dr2 − c(t, r, θ, φ)dθ2 − d(t, r, θ, φ)dφ2 .

e 3, · · · , 3) should be ds e . Then
by Schwarzschild metrics in R3 . But we know the metric in R(3,
| {z } R
m

how to we explain the differences (dsRe − dshuman ) in physics?

Notice that one can only observes the line element dshuman , i.e.,, a projection of dsRe on R3human
by the projective principle. Whence, all contributions in (dsRe − dshuman ) come from the spatial
direction not observable by human beings. In this case, it is difficult to determine the exact
behavior and sometimes only partial information of the Universe, which means that each law on
the Universe determined by human beings is an approximate result and hold with conditions.
Furthermore, if mb ≤ 2 holds, because there are infinite underlying connected graphs, i.e.,
there are infinite combinations of R3 , one can not find an ultimate theory for the Universe,
which means the discover of science for human beings will endless forever, i.e., there are no a
Theory of Everything.

References

[1] Armstrong M.A., Basic Topology, McGraw-Hill. Berkshire, England 1979.

[2] Belot G., Remarks on the geometry of visibles, Philosophical Quarterly, 55, 581-586(2003).
[3] Carmeli M., Classical Fields–General Relativity and Gauge Theory, World Scientific 2001.
[4] Chartrand G. and Lesniak L., Graphs & Digraphs, Wadsworth Inc. California 1986.
[5] Erdös P., Goodman A. and Pósa L., The representation of a graph by set intersections,
Canda. J. Math., 18, 106-112(1966).
[6] Kaku M., Parallel Worlds, Doubleday, An imprint of Random House 2004.
[7] Mao L.F. Automorphism Groups of Maps, Surfaces and Smarandache Geometries, Ameri-
can Research Press 2005.
[8] Mao L.F. Smarandache Multi-Space Theory, Hexis. Phoenix. USA 2006.
[9] Mao L.F., On multi-metric spaces, Scientia Magna, 1, 87-94(2006).
[10] Mao L.F., On algebraic multi-vector spaces, Scientia Magna, 2, 1-6(2006).
[11] Mao L.F., Combinatorial speculation and combinatorial conjecture for mathematics, In-
ternational J.Math. Combin., Vol.1, 1-19(2007).
[12] Mao L.F., Geometrical theory on combinatorial manifolds, JP J.Geometry and Topology,
1,65-114(2007).
[13] Mao L.F.,Extending homomorphism theorem to multi-systems, International J.Math. Com-
bin., 3, 1-27(2008).
[14] Massey W.S., Algebraic topology: an introduction, Springer-Verlag, New York 1977.
[15] Papantonopoulos E., Braneworld cosmological models, arXiv: gr-qc/0410032.
[16] Papantonopoulos E., Cosmology in six dimensions, arXiv: gr-qc/0601011.
64 L.F.Mao

[17] Polchinski J., String Theory, Cambridge University Press 1998.

[18] Reid T., An inquiry into the human mind on the principles of common sense, Edinburgh
University Press 1997.
[19] Smarandache F, Mixed noneuclidean geometries, arXiv: math/0010119, 10/2000.
[20] Tegmark M., Parallel Universes, in : Science and Ultimate Reality: From Quantum to
Cosmos ( Eds: Barrow J.D., Davies P.C.W. and Harper C.L.) Cambridge University Press
2003.
[21] Yaffe G., Reconsidering Reid’s geometry of visibles’, Philosophical Quarterly, 52, 602-
620(2002)
Math.Combin.Book Ser. Vol.1 (2010), 65-73

Counting Rooted Eulerian Planar Maps

Junliang Cai

Laboratory of Mathematics and Complex Systems

School of Mathematical Sciences, Beijing Normal University, Beijing 100875, China

Yanpei Liu

Department of Mathematics, Northern Jiaotong University, Beijing 100044, China

Abstract: In this paper a new method for establishing generating equations of rooted
Eulerian planar maps will be provided. It is an algebraic method instead of the constructional
one used as before and plays an important role in finding the kind of equations. Some
equations of rooted loopless Eulerian planar maps will be obtained by using the method and
some results will be corrected and simplified here.
Keywords: Eulerian map, generating function, enumerating equation, Smarandache
multi-embedding, multi-surface.
MSC(2000): 05A15, 05C30

§1. Introduction

A Smarandache multi-embedding of a graph G on a multi-surface Se is a continuous mapping

ς : G → Se such that there are no intersections between any two edges unless its endpoints,
where Se is an unions of surfaces underlying a graph H. Particularly, if |V (H)| = 1, i.e., Se is
just a surface, such multi-embedding is the common embedding of G.
With respect to the enumeration of rooted Eulerian planar maps the first result for enu-
merating rooted general Eulerian planar maps with vertex partition was achieved by Tutte [10]
in the early 1960’s. In 1986 the enumeration of rooted non-separable Eulerian planar maps with
vertex partition was studied by Liu [4]. In 1992 the enumeration of rooted loopless Eulerian
planar maps with vertex partition and other variables as parameters were investigated by Liu
[5,6,7] too. From then on some new results were obtained [1,2,8,9], but the method used there
was so difficult that one can not understand them easily. In 2004 the enumeration of unrooted
Eulerian and unicursl planar maps with the number of edges was resulted by Liskovets [3] based
on the rooted results. In present article we will provide an algebraic method instead of that
used in the past for counting this kind of planar maps. It will paly an important role in estab-
lishing the equations of all kinds of rooted Eulerian planar maps. As examples, some equations
of rooted loopless Eulerian planar maps can be derived by using the method. The procedure
1 Received March 8, 2010. Accepted March 25, 2010.
66 Junliang Cai and Yanpei Liu

and some results in [5,6,8] will be reduced greatly and updated properly.
In general, rooting a map means distinguishing one edge on the boundary of the outer face
as the root-edge, and one end of that edge as the root-vertex. In diagrams we usually represent
the root-edge as an edge with an arrow in the outer face, the arrow being drawn from the
root-vertex to the other end. So the outer face is also called the root-face. A planar map
with a rooting is said to be a rooted planar map. We say that two rooted planar maps are
combinatorially equivalent or up to root-preserving isomorphism if they are related by one to
one correspondence of their elements, which maps vertices onto vertices, edges onto edges and
faces onto faces,and which preserves incidence relations and the rooted elements. Otherwise,
combinatorially inequivalent or nonisomorphic here.
Let M be any set of maps. For a map M ∈ M let M − R and M • R be the resultant maps
of deleting the root-edge R(M ) from M and contracting R(M ) into a vertex as the new root-
vertex, respectively. For a vertex v of M let val(v) be the valency of the vertex v. Moreover,
the valency of the root-vertex of M is denoted by val(M ).
Terminologies and notations not explained here refer to [9].

§2. Relations on Maps

In order to set up the enumerating equation satisfied by some generating functions we have to
introduce the operations on maps in M.
Let
MhRi = {M − R | M ∈ M}; M(R) = {M • R | M ∈ M}, (31)

and let 
e P

▽M = {▽i M | i = 1, 2, · · · , l(M ) − 1};
M∈M
P (32)

▽M = {▽i M | i = 0, 1, 2, · · · , l(M )},
M∈M

where ▽i M is the resultant map of splitting the root-vertex of M into two vertices vr′ and vr′′
with a new edge hvr′ , vr′′ i as the root-edge of the new map ▽i M such that the valency of its
root-vertex val(▽i M ) = i + 1.
Further,write that

M(e) = {M ∈ M | val(M ) ≡ 0(mod2)};
(33)
M(o) = {M ∈ M | val(M ) ≡ 1(mod2)}.

It is clear that M(e) and M(o) stand for maps in M with the valency of root-vertex of the
maps being even and odd, respectively.
Let M1 and M2 be two sets of maps. For two maps M1 ∈ M1 and M2 ∈ M2 , let M1 +̇M2
S
be the map M1 M2 such that
T
(i) M1 M2 is only a vertex as the root-vertex of M1 +̇M2 ;
(ii) M1 is inside one of the faces incident with the root-vertex of M2 ;
(iii) The root-edge of M1 +̇M2 is the same as that of M2 ;
 Counting Rooted Eulerian Planar Maps 67

(iv) The first occurrence of the edges in M1 incident with the root-vertex of M1 +̇M2 is the
root-edge of M1 when one moves around the root-vertex of M1 +̇M2 in the rotational direction
starting from the root-edge of M1 +̇M2 .
For the maps Mi ∈ Mi , i = 1, 2, · · · , k, we define that



M1 +̇M2 +̇ · · · +̇Mk = M1+̇M2 +̇ · · · +̇Mk−1 +̇Mk ;

M1 ⊙ M2 ⊙ · · · ⊙ Mk = M1 +̇M2 +̇ · · · +̇Mk | Mi ∈ Mi , 1 6 i 6 k , (34)


 ⊙k
M = M1 ⊙ M2 ⊙ · · · ⊙ Mk |M1 =M2 =···=Mk =M .

Now, we have to introduce another kind operation in order to finish the construction of
the sets of maps as follows.
For two maps M1 ∈ M1 and M2 ∈ M2 , let M1 +̂M2 be the resultant map of identifying
the two root-edges of M1 and M2 such that M1 is inside the non-root-face incident with the
root-edge of M2 , or onto the non-root-side of M2 if the root-edge of M2 is a cut-edge. Of course,
the root-edge of M1 +̂M2 has to be the identified edge and the non-root-face incident with the
root-edge of M1 +̂M2 is the same as in M1 .
For the maps M ∈ M and Mi ∈ Mi , i = 1, 2, · · · , k, we define that



M1 +̂M2 +̂ · · · +̂Mk = M1+̂M2 +̂ · · · +̂Mk−1 +̂Mk ;

M1 ⊕ M2 ⊕ · · · ⊕ Mk = M1 +̂M2 +̂ · · · +̂Mk | Mi ∈ Mi , 1 6 i 6 k ; (35)


 ⊕k
M = M1 ⊕ M2 ⊕ · · · ⊕ Mk |M1 =M2 =···=Mk =M .

A map is called Eulerian if all its vertices are of even valency. It is well-known that a
map is Eularian if and only if it has an Eularian circuit, a circuit containing each of the edges
exactly once. A map is called loopless if there is no any loop in the map.
Let Enl be the set of all rooted loopless Eulerian planar maps with the vertex map ϑ in Enl
as a special case. Of course, the loop map O is not in Enl . It is easily checked that no Eulerian
maps has a separable edge.
The enumerating problems of rooted loopless Eulerian planar maps will be discussed here
by using a new method witch is much simpler than that used in the past [4,5,6,9].
Let Enl0 = {ϑ} and Enli = {M ∈ Enl − ϑ | R(M) is i multi-edges in M }, for i > 1. Then
the set Enl can be partitioned into the following form
X X
Enl = Enli , and Enl (R) = Enli (R), (36)
i>0 i>0

where Enl0 (R) = Enl0 = {ϑ} and Enl1 (R) = Enl − Enl0 . Let Ein = Enl +̇{O} be the set of all
rooted inner Eulerian planar maps [9], then from (4) we have

⊙i−1
Enli (R) = Ein ⊙ Enl ,

for i > 2.
68 Junliang Cai and Yanpei Liu

⊙0
If we write Ein = Enl0 = {ϑ}, then from (6) we have
X X
Enl (R) = Enli (R) = Enl0 (R) + Enl1 (R) + Enli (R)
i>0 i>2
X
⊙i−1
= Enl0 + (Enl − Enl0 ) + (Ein ⊙ Enl )
i>2
X X
⊙i ⊙i
= Enl + (Ein ⊙ Enl ) = Enl0 ⊙ Enl + (Ein ⊙ Enl )
i>1 i>1
X
⊙0 ⊙i
= Ein ⊙ Enl + (Ein ⊙ Enl ).
i>1

i.e., X
⊙i
Enl (R) = (Ein ⊙ Enl ). (37)
i>0

Now, In order to enumerate the maps in Enl conveniently, we need to reconstruct the set
Enl according to the construction of Enl (R) in (7). Hence, we suppose that
X  ⊕i 
F= e
▽Ein ⊕ (▽Enl ) , (38)
i>0
 ⊕0
where ▽Ee in is defined as Enl0 .
In general, a map in F may be not Eulerian. It is obvious that F can be classified into
two classes F (e) and F (o) where F (e) is just what we need because the maps in it are all
Eulerian, i.e., F (e) ⊆ Enl − Enl0 . Conversely, for any map M ∈ Enl − Enl0 , there is a set Enli ,
⊙i−1
i > 1 such that M ∈ Enli , thus M • R ∈ Enli (R) = Ein ⊙ Enl . So we have M ∈ Enli =
h i(e)
e in )⊕i−1 ⊕ (▽Enl )
(▽E ⊂ F (e) , i.e., Enl − Enl0 ⊆ F (e) . In the other words, we have

Enl = Enl0 + F (e) and F (e) = F − F (o) . (39)

In addition, it is not difficult to see that

Ein hRi = Enl . (40)

§3. Equations with Vertex Partition

In this section we want to discuss the following generating function for the set M of some maps.
X
gM (x : y) = xl(M) y n(M) , (41)
M∈M

in which y(M ) and n(M ) stand for infinite vectors, and

Y n (M)
y n(M) = yi i ; y = (y1 , y2 , · · · ); n(M ) = (n1 (M ), n2 (M ), · · · ),
i>1

where l(M ) = val(M ) and ni (M ) is the number of the non-root vertices of valency i, i > 1.
The function (11) is said to be the vertex partition function of M. Naturally, for a Eulerian
planar map M ∈ Enl , we may let l(M ) = val(M ) = 2m(M ) and n2j+1 (M ) ≡ 0 for j > 0.
 Counting Rooted Eulerian Planar Maps 69

For this reason we need to introduce the following Blisard -operator in y

Z Z
y i = yi , i > 1 and y 0 = 1,
y y

which is a linear operator and for a function f (z) we define that

f (x) − f (y)
δx,y f = . (42)
x2 − y 2

They are said to be (x, y)-deference of f (z) .

In the following, the new algebraic method is used for enumerating the set of maps in F (e) .

Lemma 3.1 For the set F (e) , we have

Z
x2 y 2 δx,y (f + z 2 f 2 )
gF (e) (x : y) = , (43)
¯ y 1 − x2 y 2 δx,y (2f + z 2 f 2 )

where f = f (z) = gEnl (z : y).

¯
Proof From the definitions (3), (8) and (11), we have
 i
X Z X l(M)−1
X X l(M)
X
gF (x : y) = x y xj y l(M)−j yn(M)  xj y l(M)−j y n(M)
i>0 y M∈Ein j=1 M∈Enl j=0
Z X i
gE (y) − x−1 ygEin (x) f (y) − xy −1 f (x)
= x y xy −1 in
y i>0 1 − xy −1 1 − xy −1
Z X i+1
yf (y) − xf (x)
= xy
y i>0 y−x
Z
xy(yf (y) − xf (x))
= 2 2
y y (1 + x f (x)) − x (1 + y f (y))
Z 2

2


2 2 1 + y f (y) f (y) − 1 + x f (x) f (x)
= x y + gF (o) (x : y)
y y 2 (1 + x2 f (x))2 − x2 (1 + y 2 f (y))2

i.e., Z
x2 y 2 δx,y (f + z 2 f 2 )
gF (x : y) = + gF (o) (x : y),
y 1 − x2 y 2 δx,y (2f + z 2 f 2 )
where f = f (z) = gEnl (z : y) and
Z
xyδx,y (z 2 f )
gF (o) (x : y) = . (44)
y 1 − x2 y 2 δx,y (2f + z 2 f 2 )

This lemma can be derived from (9) immediately. 

Theorem 3.1 The generating function f = f (z) = gEnl (z : y) with vertex partition satisfies
¯
the following enumerating equation
Z
1 − x2 y 2 δx,y f
f= 2 2 2 2
, (45)
y 1 − x y δx,y (2f + z f )
70 Junliang Cai and Yanpei Liu

This is a modification and simplification to the result (3.13) in [5].

Proof It is clear that gEnl0 (x : y) = 1. So from (9) and (13), Eq.(15) is obtained by
grouping the terms. 

§4. Equations with the Numbers of Vertices and Faces

In what following we want to study the following generating function for the set M of some
maps.
X
fM (x, y, z) = xl(M) y n(M) z q(M) , (46)
M∈M

where l(M ) = val(M ) and n(M ) and q(M ) are the numbers of non-root vertices and inner faces
of M ∈ M, respectively. It is clear that we may write l(M ) = val(M ) = 2m(M ) if M ∈ Enl is
an Eulerian map.
In fact, this section will provide a functional equation satisfied by the generating function
f = fEnl (x, y, z) with the valency of root-vertex, the numbers of non-root vertices and inner
faces of the maps in Enl , respectively, as three parameters.
Summing the results as above, we can obtain the following results.

Lemma 4.1 For the set Ein , we have

fEin (x, y, z) = x2 zf, (47)

where f = fEnl (x, y, z).

Proof The Lemma is obtained directly from (10) and (16). 

In the following, the algebraic method is used again for enumerating the set of maps in
(e)
F .

Lemma 4.2 For the set F (e) , we have

2 (1 + zf ∗ ) f ∗ − 1 + x2 zf f
fF (e) (x, y, z) = x y 2 2, (48)
(1 + x2 zf ) − x2 (1 + zf ∗ )

where f = fEnl (x, y, z) and f ∗ = fEnl (1, y, z).

Proof By (8),(9),(12) and (16) we have

 i
2m(M)−1 2m(M)
X X X X X
fF (x, y, z) = xy  xj y n(M) z q(M)  xj y n(M) z q(M)
i>0 M∈Ein j=1 M∈Enl j=0

X xfE∗in − fEin
i
f ∗ − xf X f ∗ − xf
i+1
= xy = yz −1 xz
1−x 1−x 1−x
i>0 i>0
∗
xy(f − xf )
= ,
1 − x (1 + zf ∗ ) + x2 zf
 Counting Rooted Eulerian Planar Maps 71

i.e.,

(1 + zf ∗ ) f ∗ − 1 + x2 zf f
fF (x, y, z) = x2 y 2 2 + fF (o) (x, y, z),
(1 + x2 zf ) − x2 (1 + zf ∗ )
where fE∗in = fEin (1, y, z) and


xy f ∗ − x2 f
fF (o) (x, y, z) = 2 2. (49)
(1 + x2 zf ) − x2 (1 + zf ∗ )

The Lemma is obtained directly from the definition of F (e) in (9). 

Theorem 4.1 The generating function f = fEnl (x, y, z) with the valency of root-vertex, the
numbers of non-root vertices and inner faces of the maps in Enl , respectively, as three parameters
satisfies the following cubic equation
∗ ∗ 2


2 (1 + zf ) f − 1 + x zf f
f =1+x y 2 2, (50)
(1 + x2 zf ) − x2 (1 + zf ∗ )
where f ∗ = f (1, y, z).

Proof From (9) we have

f = fEnl0 (x, y, z) + fF (e) (x, y, z),

where fEnl0 (x, y, z) = 1. By substituting (18) into the above formula Eq(20) holds. 

§5. Equations with the Edge Number and the Root-Face Valency

In this section we study the following generating function for the set M of some maps.
X
fM (x, y, z) = xl(M) y s(M) z p(M) , (51)
M∈M

where l(M ) = val(M ) and s(M ) and p(M ) are the number of edges and the valency of root-face
of M ∈ M, respectively. we may also write l(M ) = val(M ) = 2m(M ) if M ∈ Enl is an Eulerian
map.
In this section we provide a functional equation satisfied by the generating function f =
fEnl (x, y, z) with the valency of root-vertex, the number of edges the valency of the root-face of
the maps in Enl , respectively, as three parameters. Write that

hEnl (x, y) = fEnl (x, y, 1), FEnl (y, z) = fEnl (1, y, z), HEnl (y) = fEnl (1, y, 1).

Lemma 5.1 For the set Ein , we have

fEin (x, y, z) = x2 yzf, (52)

where f = fEnl (x, y, z).

Proof The Lemma is obtained directly from (10) and (21). 

72 Junliang Cai and Yanpei Liu

Lemma 5.2 For the set F (e) , we have

x2 yz[F H0 − (1 + x2 yh)f ] (1 − z)x2 y 2 zHF hf
fF (e) (x, y, z) = − , (53)
(1 + x2 yh)2 − x2 H02 1 − x2 y 2 H 2 h2
where h = hEnl (x, y), F = FEnl (y, z), H = HEnl (y) and H0 = 1 + yH.

Proof By (8),(9),(12) and (21) we have

 i
2m(M)−1 2m(M)
X X X X X
fF (x, y, z) = xyz  xj y s(M)  xj y s(M) z p(M)
i>0 M∈Ein j=1 M∈Enl j=0
X
− xk y k (z − z 2 )f hk−1 H k−1 F
k>1
X  xHE − hE i F − xf X
= xyz in in
− xyz(1 − z)F f (xyHh)k−1
1−x 1−x
i>0 k>1
xyz(F − xf ) (1 − z)xyzF f
= − ,
1 − x − xHEin + hEin 1 − xyHh

where HEin = yH,hEin = x2 yh, i.e.,

x2 yz[F H0 − (1 + x2 yh)f ] (1 − z)x2 y 2 zHF hf

fF (x, y, z) = − + fF (o) (x, y, z),
(1 + x2 yh)2 − x2 H02 1 − x2 y 2 H 2 h2
where H0 = 1 + yH and
xyz[(1 + x2 yh)F − x2 H0 f ] (1 − z)xyzF f
fF (o) (x, y, z) = − . (54)
(1 + x2 yh)2 − x2 H02 1 − x2 y 2 H 2 h2

The Lemma is obtained directly from the definition of F (e) in (9). 

Theorem 5.1 The generating function f = fEnl (x, y, z) with the valency of root-vertex, the
numbers of non-root vertices and inner faces of the maps in Enl , respectively, as three parameters
satisfies the following cubic equation
x2 yz[H0 F − (1 + x2 yh)f ] (1 − z)x2 y 2 zHF hf
f =1+ − , (55)
(1 + x2 yh)2 − x2 H02 1 − x2 y 2 H 2 h2
where h = hEnl (x, y), F = FEnl (y, z), H = HEnl (y) and H0 = 1 + yH.

Proof From (9) we have

f = fEnl0 (x, y, z) + fF (e) (x, y, z),

where fEnl0 (x, y, z) = 1. By substituting (23) into the above formula Eq(25) holds. 

Theorem 5.2 The generating function h = hEnl (x, y) with the valency of root-vertex and the
number of edges of the maps in Enl , respectively, as two parameters satisfies the following cubic
equation
h30 − h20 − (y + H02 )x2 h0 + x2 H02 + x4 yH0 = 0, (56)
 Counting Rooted Eulerian Planar Maps 73

where H0 = 1 + yHEnl (y) and h0 = 1 + x2 yhEnl (x, y).

This is a modification and simplification to the result (4.11) in [5].

Proof For any map M ∈ Enl , since the number of vertices of M is n(M ) + 1 and the
number of faces of M is q(M ) + 1, the number s(M ) of edges of M is n(M ) + q(M ) by Eulerian
formula. It follows from (16) and (21) that h = hEnl (x, y) = fEnl (x, y, y). So if we take z = y,
then Eq(20) becomes Eq(26) by grouping the terms where H = fE∗nl (y, y) = fEnl (1, y, y).
Of course, Eq(26) may be also derived by substituting y2i = y i into Eq(15) and replacing
P
x in it with x2 y since s(M ) = i>0 in2i (M ), or by substituting z = 1 into Eq(25).
2

Note that Eq(20) and Eq(26) have been solved in the forms of parametric expressions or
explicit formulae in [2] and [7], respectively.

References

[1] Cai J L, Hao R X, Liu Y P. Enumerating rooted Eulerian planar maps[J]. Acta Mathematica
Scientia 2001,21:289.
[2] Cai J L. The number of rooted loopless Eulerian planar maps(Chinese)[J]. Acta Math.
Appl.. 2006,29(2):210.
[3] Liskocets V A, Walsh T R S. Enumeration of Eulerian and unicursal planar maps[J].
Discrete Mathematics. 2004,282:209.
[4] Liu Y P. A functional equation for enumerating non-separable Eulerian planar maps with
vertex partition[J]. Kexue Tongbao(Chinese) 1986,31:81.
[5] Liu Y P. On the vertex partition equation of loopless Eulerian planar maps[J]. Acta Math.
Appl. Sinica. 1992,8:45.
[6] Liu Y P. A note on the number of loopless Eulerian planar maps[J]. J. Math. Res. Expos..
1992,12:165.
[7] Liu Y P. On the number of Eulerian planar maps[J]. Acta Math. Sinica.1992,12:418.
[8] Liu Y P. On functional equations arising from map enumerations[J]. Discrete Math. 1993,123:93.
[9] Liu Y P. Enumerative Theory of maps[M]. Kluwer, Boston. 1998.
[10] Tutte W T. A census of slicings[J]. Canad. J. Math.. 1962,14:708.
Math.Combin.Book Ser. Vol.1 (2010), 74-79

The nth Power Signed Graphs-II

P. Siva Kota Reddy† , S. Vijay†† and V. Lokesha‡

†
Department of Mathematics, Rajeev Institute of Technology, Industrial Area,

B-M Bypass Road, Hassan 573 201, India

††
Department of Mathematics, Govt. First Grade College, Kadur, Chikkamangalore 577 548, India

‡
Department of Mathematics, Acharaya Institute of Technology, Soldevanahalli, Bangalore-90, India
† ††
E-mail: reddy− [email protected]; vijays [email protected]; ‡ [email protected]

Abstract: A Smarandachely k-signed graph (Smarandachely k-marked graph) is an ordered

pair S = (G, σ) (S = (G, µ)) where G = (V, E) is a graph called underlying graph of S and
σ : E → (e1 , e2 , ..., ek ) (µ : V → (e1 , e2 , ..., ek )) is a function, where each ei ∈ {+, −}.
Particularly, a Smarandachely 2-signed graph or Smarandachely 2-marked graph is called
abbreviated a signed graph or a marked graph. In this paper, we present solutions of some
signed graph switching equations involving the line signed graph, complement and nth power
signed graph operations.

Keywords: Smarandachely k-signed graphs, Smarandachely k-marked graphs, signed

graphs, marked graphs, balance, switching, line signed graph, complementary signed graph,
nth power signed graph.

AMS(2010): 05C 22

§1. Introduction

For standard terminology and notion in graph theory we refer the reader to Harary [6]; the
non-standard will be given in this paper as and when required. We treat only finite simple
graphs without self loops and isolates.
A Smarandachely k-signed graph (Smarandachely k-marked graph) is an ordered pair S =
(G, σ) (S = (G, µ)) where G = (V, E) is a graph called underlying graph of S and σ : E →
(e1 , e2 , ..., ek ) (µ : V → (e1 , e2 , ..., ek )) is a function, where each ei ∈ {+, −}. Particularly, a
Smarandachely 2-signed graph or Smarandachely 2-marked graph is called abbreviated a signed
graph or a marked graph. A signed graph S = (G, σ) is balanced if every cycle in S has an even
number of negative edges (See [7]). Equivalently a signed graph is balanced if product of signs
of the edges on every cycle of S is positive.

1 Received February 20, 2010. Accepted March 26, 2010.

 The nth Power Signed Graphs-II 75

A marking of S is a function µ : V (G) → {+, −}; A signed graph S together with a

marking µ by Sµ .
The following characterization of balanced signed graphs is well known.

Proposition 1.1(E. Sampathkumar [8]) A signed graph S = (G, σ) is balanced if, and only if,
there exist a marking µ of its vertices such that each edge uv in S satisfies σ(uv) = µ(u)µ(v).

Given a marking µ of S, by switching S with respect to µ we mean reversing the sign of

every edge of S whenever the end vertices have opposite signs in Sµ [1]. We denote the signed
graph obtained in this way is denoted by Sµ (S) and this signed graph is called the µ-switched
signed graph or just switched signed graph. A signed graph S1 switches to a signed graph S2
(that is, they are switching equivalent to each other), written S1 ∼ S2 , whenever there exists a
marking µ such that Sµ (S1 ) ∼= S2 .
Two signed graphs S1 = (G, σ) and S2 = (G′ , σ ′ ) are said to be weakly isomorphic (see
[13]) or cycle isomorphic (see [14]) if there exists an isomorphism φ : G → G′ such that the
sign of every cycle Z in S1 equals to the sign of φ(Z) in S2 . The following result is well known
(See [14]):

Proposition 1.2(T. Zaslavsky [14]) Two signed graphs S1 and S2 with the same underlying
graph are switching equivalent if, and only if, they are cycle isomorphic.

Behzad and Chartrand [4] introduced the notion of line signed graph L(S) of a given signed
graph S as follows: Given a signed graph S = (G, σ) its line signed graph L(S) = (L(G), σ ′ )
is the signed graph whose underlying graph is L(G), the line graph of G, where for any edge
ei ej in L(S), σ ′ (ei ej ) is negative if, and only if, both ei and ej are adjacent negative edges in
S. Another notion of line signed graph introduced in [5], is as follows:

The line signed graph of a signed graph S = (G, σ) is a signed graph L(S) = (L(G), σ ′ ),
where for any edge ee′ in L(S), σ ′ (ee′ ) = σ(e)σ(e′ ). In this paper, we follow the notion of line
signed graph defined by M. K. Gill [5] (See also E. Sampathkumar et al. [9]).

Proposition 1.3(M. Acharya [2]) For any signed graph S = (G, σ), its line signed graph
L(S) = (L(G), σ ′ ) is balanced.

For any positive integer k, the k th iterated line signed graph, Lk (S) of S is defined as
follows:

L0 (S) = S, Lk (S) = L(Lk−1 (S)).

Corollary 1.4 For any signed graph S = (G, σ) and for any positive integer k, Lk (S) is
balanced.

Let S = (G, σ) be a signed graph. Consider the marking µ on vertices of S defined as

follows: for each vertex v ∈ V , µ(v) is the product of the signs on the edges incident with
v. The complement of S is a signed graph S = (G, σ c ), where for any edge e = uv ∈ G,
76 P. Siva Kota Reddy, S. Vijay and V. Lokesha

σ c (uv) = µ(u)µ(v). Clearly, S as defined here is a balanced signed graph due to Proposition
1.1.

§2. nth Power signed graph

The nth power graph Gn of G is defined in [3] as follows:

The nth power has same vertex set as G, and has two vertices u and v adjacent if their
distance in G is n or less.

In [12], we introduced a natural extension of the notion of nth power graphs to the realm
of signed graphs: Consider the marking µ on vertices of S defined as follows: for each vertex
v ∈ V , µ(v) is the product of the signs on the edges incident at v. The nth power signed graph
of S is a signed graph S n = (Gn , σ ′ ), where Gn is the underlying graph of S n , where for any
edge e = uv ∈ Gn , σ ′ (uv) = µ(u)µ(v).
The following result indicates the limitations of the notion of nth power signed graphs as
introduced above, since the entire class of unbalanced signed graphs is forbidden to nth power
signed graphs.

proposition 2.1(P. Siva Kota Reddy et al.[12]) For any signed graph S = (G, σ), its nth power
signed graph S n is balanced.

For any positive integer k, the k th iterated nth power signed graph, (S n )k of S is defined
as follows:

(S n )0 = S, (S n )k = S n ((S n )k−1 ).

Corollary 2.2 For any signed graph S = (G, σ) and any positive integer k, (S n )k is balanced.

The degree of a signed graph switching equation is then the maximum number of operations
on either side of an equation in standard form. For example, the degree of the equation S ∼ L(S)
is one, since in standard form it is L(S) ∼ S, and there is one operation on each side of the
equation. In [12], the following signed graph switching equations are solved:

• S ∼ (L(S))n (1)

• L(S) ∼ (L(S))n (2)

n
• L(S) ∼ S , where n ≥ 2 (3)

• L2 (S) ∼ S n , where n ≥ 2 (4)

• L2 (S) ∼ S n , where n ≥ 2, and (5)

n
• L2 (S) ∼ S , where n ≥ 2. (6)

Recall that L2 (S) is the second iterated line signed graph S.

 The nth Power Signed Graphs-II 77

Several of these signed graph switching equations can be viewed as generalized of earlier
work [11]. For example, equation (1) is a generalization of L(S) ∼ S, which was solved by Siva
Kota Reddy and Subramanya [11]. When n = 1 in equations (3) and (4), we get L(S) ∼ S
and L2 (S) ∼ S 2 , which was solved in [11]. If n = 1 in (5) and (6), the resulting signed graph
switching equation was solved by Siva Kota Reddy and Subramanya [11].
Further, in this paper we shall solve the following three signed graph switching equations:

• L(S) ∼ S n (7)

• L(S) ∼ S n (orL(S) ∼ S n ) (8)

• L(S) ∼ (S)n (9)

In the above expressions, the equivalence (i.e, ∼) means the switching equivalent between
corresponding graphs.
Note that for n = 1, the equation (7) is reduced to the following result of E. Sampathkumar
et al. [10].

Proposition 2.3(E. Sampathkumar et al. [10]) For any signed graph S = (G, σ), L(S) ∼ S
if, and only if, S is a balanced signed graph and G is 2-regular.

Note that for n = 1, the equations (8) and (9) are reduced to the signed graph switching
equation which is solved by Siva Kota Reddy and Subramanya [11].

Proposition 2.4 (P. Siva Kota Reddy and M. S. Subramanya [11]) For any signed graph
S = (G, σ), L(S) ∼ S if, and only if, G is either C5 or K3 ◦ K1 .

§3. The Solution of L(S) ∼ S n

We now characterize signed graphs whose line signed graphs and its nth power line signed
graphs are switching equivalent. In the case of graphs the following result is due to J. Akiyama
et. al [3].

Proposition 3.1(J. Akiyama et al. [3]) For any n ≥ 2, the solutions to the equation L(G) ∼
= Gn
are graphs G = mK3 , where m is an arbitrary integer.

Proposition 3.2 For any signed graph S = (G, σ), L(S) ∼ S n , where n ≥ 2 if, and only if, G
is mK3 , where m is an arbitrary integer.

Proof Suppose L(S) ∼ S n . This implies, L(G) ∼

= Gn and hence by Proposition 3.1, we see
that the graph G must be isomorphic to mK3 .
Conversely, suppose that G is mK3 . Then L(G) ∼ = Gn by Proposition 3.1. Now, if S is
a signed graph with underlying graph as mK3 , by Propositions 1.3 and 2.1, L(S) and S n are
balanced and hence, the result follows from Proposition 1.2. 
78 P. Siva Kota Reddy, S. Vijay and V. Lokesha

§4. Solutions of L(S) ∼ S n

In the case of graphs the following result is due to J. Akiyama et al. [3].

Proposition 4.1(J. Akiyama et al. [3]) For any n ≥ 2, G = C2n+3 is the only solution to the
equation L(G) ∼
= Gn .

Proposition 4.2 For any signed graph S = (G, σ), L(S) ∼ S n , where n ≥ 2 if, and only if, G
is C2n+3 .
∼ Gn and hence by Proposition 4.1, we see
Proof Suppose L(S) ∼ S n . This implies, L(G) =
that the graph G must be isomorphic to C2n+3 .
Conversely, suppose that G is C2n+3 . Then L(G) ∼ = Gn by Proposition 4.1. Now, if S is
a signed graph with underlying graph as C2n+3 , by definition of complementary signed graph
and Proposition 2.1, L(S) and S n are balanced and hence, the result follows from Proposition
1.2. 
∼ (G) , n ≥ 2. So
In [3], the authors proved there are no solutions to the equation L(G) = n

its very difficult, in fact, impossible to construct switching equivalence relation of L(S) ∼ (S)n .

References

[1] R. P. Abelson and M. J. Rosenberg, Symoblic psychologic:A model of attitudinal cognition,

Behav. Sci., 3 (1958), 1-13.
[2] M. Acharya, x-Line sigraph of a sigraph, J. Combin. Math. Combin. Comput., 69(2009),
103-111.
[3] J. Akiyama, K. Kanoko and S. Simic, Graph equations for line graphs and n-th power
graphs I, Publ. Inst. Math. (Beograd), 23 (37) (1978), 5-8.
[4] M. Behzad and G. T. Chartrand, Line-coloring of signed graphs, Elemente der Mathematik,
24(3) (1969), 49-52.
[5] M. K. Gill, Contributions to some topics in graph theory and its applications, Ph.D. thesis,
The Indian Institute of Technology, Bombay, 1983.
[6] F. Harary, Graph Theory, Addison-Wesley Publishing Co., 1969.
[7] F. Harary, On the notion of balance of a signed graph, Michigan Math. J., 2(1953), 143-146.
[8] E. Sampathkumar, Point signed and line signed graphs, Nat. Acad. Sci. Letters, 7(3)
(1984), 91-93.
[9] E. Sampathkumar, P. Siva Kota Reddy, and M. S. Subramanya, The Line n-sigraph of a
symmetric n-sigraph, Southeast Asian Bull. Math., to appear.
[10] E. Sampathkumar, P. Siva Kota Reddy, and M. S. Subramanya, Common-edge signed
graph of a signed graph-Submitted.
[11] P. Siva Kota Reddy, and M. S. Subramanya, Signed Graph Equation LK (S) ∼ S, Inter-
national J. Math. Combin., 4 (2009), 84-88.
[12] P. Siva Kota Reddy, S. Vijay and V. Lokesha, nth Power signed graphs, Proceedings of the
Jangjeon Math. Soc., 12(3) (2009), 307-313.
 The nth Power Signed Graphs-II 79

[13] T. Sozánsky, Enueration of weak isomorphism classes of signed graphs, J. Graph Theory,
4(2)(1980), 127-144.
[14] T. Zaslavsky, Signed Graphs, Discrete Appl. Math., 4(1)(1982), 47-74.
Math.Combin.Book Ser. Vol.1 (2010), 80-86

Dynamical Knot and Their Fundamental Group

M. Abu-Saleem

(Department of Mathematics, Al-Laith college for girls,Umm AL-Qura University, Saudi Arabia)

E-mail: [email protected]

Abstract: In this article, we introduce the fundamental group of the dynamical trefoil
knot. Also the fundamental group of the limit dynamical trefoil knot will be achieved. Some
types of conditional dynamical manifold restricted on the elements of a free group and their
fundamental groups are presented. The dynamical trefoil knot of variation curvature and
torsion of manifolds on their fundamental group are deduced .Theorems governing these
relations are obtained.

Keywords: Dynamical trefoil knot, fundamental group, knot group, Smarandache multi-
space.

AMS(2010): 51H20, 57N10, 57M05, 14F35, 20F34

§1. Introduction

A means of describing how one state develops into another state over the course of time.
Technically, a dynamical system is a smooth action of the reals or the integers on another object
(usually a manifold). When the reals are acting, the system is called a continuous dynamical
system, and when the integers are acting, the system is called a discrete dynamical system. If
f is any continuous function, then the evolution of a variable x can be given by the formula
xn+1 = f (xn ). This equation can also be viewed as a difference equation xn+1 −xn = f (xn )−xn ,
so defining g(x) ≡ f (x) − x gives xn+1 − xn = g(xn ) ∗ 1,which can be read ”as n changes by 1
unit, x changes by g(x). This is the discrete analog of the differential equation x′ (n) = g(x(n)).
In other words; a dynamic system is a set of equations specifying how certain variables
change over time. The equations specify how to determine (compute) the new values as a
function of their current values and control parameters. The functions, when explicit, are
either difference equations or differential equations. Dynamic systems may be stochastic or
deterministic. In a stochastic system, new values come from a probability distribution. In a
deterministic system, a single new value is associated with any current value [1, 11] .
The dynamical systems were discussed in [1, 9, 11]. The fundamental groups of some types
of a manifold were studied in [2, 6 − 8, 10].

1 Received December 12, 2009. Accepted March 26, 2010.

 Dynamical Knot and Their Fundamental Group 81

§1. Definitions

1. The set of homotopy classes of loops based at the point x0 with the product operation
[f ][g] = [f.g] is called the fundamental group and denoted by π1 (X, x0 ) [3].

2. Given spaces X and Y with chosen points x0 ∈ X and y0 ∈ Y , then the wedge sum X ∨ Y
is the quotient of the disjoint union X ∪ Y obtained by identifying x0 and y0 to a single
point [5].

3. A knot is a subset of 3-space that is homeomorphic to the unit circle and a trefoil knot is
the simplest nontrivial knot, it can be obtained by joining the loose ends of an overhand
knot [5].

4. A Smarandache multi-space is a union of n spaces equipped with some different structures

for an integer n ≥ 2 , which can be used for discrete or connected space [4].

5. Given a knot k, the fundamental group π1 (R3 − k) is called the knot group of k [5].

6. A dynamical system in the space X is a function q = f (p, t) which assigns to each point
p of the space X and to each real numbert , ∞ < t < ∞ a definite point q ∈ X and
possesses the following three properties :

a- Initial condition : f (p, 0) = p for any pointp ∈ X.

b- Property of continuity in both arguments simultaneously:

lim f (p, t) = f (p0 , t0 )

p→p0
t→t0

c- Group property f (f (p, t1 ), t2 ) = f (p, t1 + t2 )[11].

§2. The Main Results

Aiming to our study, we will introduce the following:

Theorem 3.1 Let K be a trefoil knot then there are two types of dynamical trefoil knot
Di : K → K , i = 1, 2 , Di (K) 6= K,which induces dynamical trefoil knot D̄i : π1 (K) → π1 (K)
such that D̄i (π1 (K)) is a free group of rank ≤ 4 or identity group.

Proof Let D1 : K → K be a dynamical trefoil knot such that D1 (K) is dynamical crossing
i.e. the point of upper arc crossing touch the point of lower crossing, where D1 (c) = p1 as in
Figure 1(a) then we have the induced dynamical trefoil knot D̄1 : π1 (K) → π1 (K) such that
D̄1 (π1 (K)) = π1 (D1 (K)) ≈ π1 (S11 ) ∗ π1 (S21 ), thus D̄1 (π1 (K)) ≈ Z ∗ Z , so D̄1 (π1 (K))is a free
group of rank = 2. Also, if D1 : K → K such that D1 (c) = p1 , D1 (b) = p2 then D1 (K) is
space as in Figure 1(b) and so D̄1 (π1 (K)) = π1 (D1 (K)) ≈ π1 (S11 ) ∗ π1 (S21 ) ∗ π1 (S31 ), thus
D̄1 (π1 (K)) is a free group of rank = 3. Moreover, if D1 : K → K such that D1 (c) =
p1 , D1 (b) = p2 , D1 (a) = p3 , then D1 (K) is space as in Figure 1(c) and so D̄1 (π1 (K)) =
82 M. Abu-Saleem

π1 (D1 (K)) ≈ π1 (S11 ) ∗ π1 (S21 ) ∗ π1 (S31 ) ∗ π1 (S41 ), hence D̄1 (π1 (K))is a free group of rank = 4.
There is another type D2 : K → K such that D2 (K) is dynamical trefoil knot with singularity
as in Figure 1(d) then we obtain the induced dynamical trefoil knot D̄2 : π1 (K) → π1 (K)
such that D̄2 (π1 (K)) = π1 (D2 (K)) = 0. Therefore, D̄i (π1 (K)) is a free group of rank≤ 4 or
identity group. 

-
p1
D1
(a)

D1 - p2 p1
(b)

-
p2 p1
D1
(c)
p3

D2 - (d)

FIGURE 1

Theorem 3.2 The fundamental group of the limit dynamical trefoil knot is the identity group.

Proof Let D1 : K → K1 , D2 : D1 (K) → D1 (K2 ) , . . . , Dn : Dn−1 (Dn−2 ) . . . (D1 (K) →

Dn−1 (Dn−2 ) . . . (D1 (Kn ) such that lim (Dn (Dn−1 ) . . . (D1 (K) . . .) is a point as in Figure 2
n→∞
(a,b) ,then π1( lim (Dn (Dn−1 ) . . . (D1 (K) . . .)) = 0. 
n→∞

Theorem 3.3 There are different types of dynamical link graph L which represent a trefoil
knot ,where D(L) 6= L such that π1 (D(L)) is a free group of rank ≤ 3.

Proof Let L be a link graph which represent a trefoil knot and consider the following
dynamical edges D(e) = a, D(f ) = c, D(g) = b as in Figure 3(a) then π1 (D(L)) ≈ π1 (S 1 ) and
so π1 (D(L)) is a free group of rank 1. Now, if D(e) 6= e, D(f ) 6= f, D(g) 6= g as in Figure
3(b)we get the same result. Also, if D(e) = e, D(f ) = f, D(g) 6= g as in Figure 3(c) then,
π1 (D(L)) ≈ π1 (S11 ) ∗ π1 (S21 ) ∗ π1 (S31 ), thus π1 (D(L)) is a free group of rank 3.Moreover, if
 Dynamical Knot and Their Fundamental Group 83

D(e) = e, D(f ) 6= f, D(g) 6= g as in Figure 3(d)then π1 (D(L)) ≈ π1 (S11 ) ∗ π1 (S21 ) .Hence

π1 (D(L)) is a free group of rank 2. Therefore π1 (D(L)) is a free group of rank ≤ 3. 

- - -
lim Dn
D1 D2 n→∞

- - -
lim Dn
D1 D2 n→∞

FIGURE 2

g
g
e a b D(L) - a b
(or)
D(L) - e
a b

c c c
f
f
(a) (b)

g
g
- -
e e
D(L) D(L)
(or) a b (or) a b

c c
f f
(c) (d)

FIGURE 3

Theorem 3.4 The fundamental group of limit dynamical link graph of n vertices is a free group
of rank n.

Proof Let K be link graph of n vertices , then lim (D(K)) is a graph with only one vertex
n→∞
84 M. Abu-Saleem

n
and n-loops as in Figure 4 ,for n=3 and so π1 ( lim (D(K))) = π1 ( ∨ Si1 ) ≈ Z
| ∗ Z {z
∗ ... ∗ Z}.
n→∞ i =1
n terms
Hence, π1 ( lim (D(K))) is a free group of rank n. 
n→∞

D1 - D2 - n→∞ -
lim Dn

FIGURE 4

Theorem 3.5 Let I be the closed interval [0, 1] . Then there is a sequence of dynamical man-
ifolds Di : I → Ii , i = 1, 2, ..., n with variation curvature and torsion such that lim Dn (I) is
n→∞
trefoil knot and π1 (R3 − lim Dn (I)) ≈ Z.
n→∞

Proof Consider the sequence of dynamical manifolds with variation curvature and torsion
: D1 : I → I1 , D2 : I1 → I2 , ..., Dn : In−1 → In such that lim Dn (I) is a trefoil knot as in
n→∞
Figure 5, Therefore, π1 (R3 − lim Dn (I)) ≈ Z. 
n→∞

D1 -
0 1

- -
lim Dn
D2 n→∞
.......

FIGURE 5

Theorem 3.6 The knot group of the limit dynamical sheeted trefoil knot is either isomorphic
to Z or identity group.

Proof Let K be a sheet trefoil knot with boundary {A, B}as in Figure 6 and D : K → K
is dynamical sheeted trefoil knot of K into itself,then we get the following sequence: D1 : K →
K, D2 : D1 (K) → D1 (K) , . . . ,
Dn : (Dn−1 ) . . . (D1 (K) . . .) → (Dn−1 ) . . . (D1 (K) . . .)
such that lim (Dn (Dn−1 ) . . . (D1 (K) . . .) = k where, k is a trefoil knot as in Figure 6(a) then
n→∞
π1( R3 −k) ≈ Z..Also, if lim (Dn (Dn−1 ) . . . (D1 (K) . . .) =point as in Figure 6(b,c) thenπ1( R3 −
n→∞
lim (Dn (Dn−1 ) . . . (D1 (K) . . .)) = π1( R3 − one point). Hence,
n→∞
 Dynamical Knot and Their Fundamental Group 85

π1( R3 − lim (Dn (Dn−1 ) . . . (D1 (K) . . .)) = 0.

n→∞

Therefore, the knot group of the limit dynamical sheeted trefoil knot is either isomorphic to
Z or identity group.

-
D1 D2- n→∞ -
lim Dn
(a)

D1
- D2
- lim Dn
n→∞
......... - (b)

- - -
lim Dn
D1 D2
.....n→∞ (c)

FIGURE 6

References

[1] G.D. Birkhoof, Dynamical systems, Bulletin of the American Mathematics Society, Vol.37,
No.1 (2000) 88-121.
[2] R.Frigerio, Hyperbolic manifold with geodesic boundary which are determine by their
fundamental group, Topology and its Application, 45 (2004) 69-81.
[3] A. Hatcher, Algebraic topology, The web addressis :
http:///www.math.coronell.edu/hatcher.
[4] Linfan Mao, Smarandache Multi-Space Theory, Hexis, Phoenix, American 2006.
[5] W. S. Massey, Algebraic Topology: An introduction, Harcourt Brace and world, New York
(1967).
[6] O. Neto and P. C. Silva, The fundamental group of an algebraic link, C. R. Acad Sci.
Paris, Ser. I, 340 (2005) 141-146.
[7] M. Abu-Saleem, Folding on the wedge sum of graphs and their fundamental group, Applied
Sceinces, 12 (2010) 14-19.
[8] M. Abu-Saleem, Folding on the chaotic Cartesian product of manifolds and their funda-
mental group, Tamakang journal of Mathematics, 39 (2008) 353-361
[9] M. Abu-Saleem, Dynamical manifold and their fundamental group, Ars Combinatoria, to
appear.
86 M. Abu-Saleem

[10] M. Abu-Saleem, Folding and unfolding of manifolds and their fundamental groups, Int.J.
Contemp. Math. Sciences, Vol.5, No.1 (2010),1-19.
[11] K. S. Sibirsky, Introduction To Topological Dynamics, Noordhoff Int. Pub. Leyden, The
Netherlands (1975)
Math.Combin.Book Ser. Vol.1 (2010), 87-98

The Crossing Number of the Cartesian Product of

Star Sn with a 6-Vertex Graph

Jing Wang1 , Yuanqiu Huang2

1. Department of Mathematics and Information Sciences, Changsha University, Changsha 410003, China

2. Department of Mathematics, Normal University of Hunan, Changsha 410081, China

E-mail: [email protected]
Abstract: Calculating the crossing number of a given graph is in general an elusive
problem and only the crossing numbers of few families of graphs are known. Most of them
are the Cartesian product of special graphs. This paper determines the crossing number of
the Cartesian product of star Sn with a 6-vertex graph.

Keywords: Smarandache P-drawing, crossing number, Cartesian product, Star.

AMS(2010): 05C, 05C62

§1. Introduction

For definitions not explained here, readers are referred to [1]. Let G be a simple graph with
vertex set V and edge set E. By a drawing of G on the plane Π, we mean a collection of points
P in Π and open arcs A in Π − P for which there are correspondences between V and P and
between E and A such that the vertices of an edge correspond to the endpoints of the open
arcs. A drawing is called good, if for all arcs in A, no two with a common endpoint meet, no
two meet in more than one point, and no three have a common point. A crossing in a good
drawing is a point of intersection of two arcs in A. A Smarandache P-drawing of a graph G for
a graphical property P is such a good drawing of G on the plane with minimal intersections
for its each subgraph H ∈ P. A Smarandache P-drawing is said to be optimal if P = G and
it minimizes the number of crossings. The crossing number cr(G) of a graph G is the number
of crossings in any optimal drawing of G in the plane. Let D be a good drawing of the graph
G, we denote by cr(D) the number of crossings in D.
Let Pn and Cn be the path and cycle of length n, respectively, and the star Sn be the
complete bipartite graph K1,n .
Given two vertex disjoint graphs G1 and G2 , the Cartesian product G1 × G2 of G1 and G2

1 Supported by National Science Foundation of China (10771062), New Century Excellent Talents in Uni-

versity (07-0276), Changsha Science and Technology Program(K0902210-11) and Talent Introduction Research
Fund(SF0905).
2 Received January 11, 2010. Accepted March 28, 2010.
88 Jing Wang and Yuanqiu Huang

is defined by


 V (G1 × G2 ) = V (G1 ) × V (G2 )





 E(G1 × G2 ) = {(u1 , u2 ) (v1 , v2 ) | u1 = v1 and u2 v2 ∈ E(G2 ),




or u2 = v2 and u1 v1 ∈ E(G1 )}

Let G1 be a graph homeomorphic to G2 , then cr(G1 ) = cr(G2 ). And if G1 is a subgraph

of G2 , it is easy to see that cr(G1 ) ≤ cr(G2 ).
Calculating the crossing number of a given graph is in general an elusive problem [2] and
only the crossing numbers of few families of graphs are known. Most of them are Cartesian
products of special graphs, partly because of the richness of their repetitive patterns. The
already known results on the crossing number of G × H fit into three categories:
(i) G and H are two small graphs. Harary, et al. obtained the crossing number of C3 × C3
in 1973 [3]; Dean and Richter [4] investigated the crossing number of C4 × C4 ; Richter and
Thomassen [5] determined the crossing number of C5 × C5 ; in [6] Anderson, et al. obtained the
crossing number of C6 × C6 ; Kles̆c̆ [7]studied the crossing number of K2,3 × C3 . These results
are usually used as the induction basis for establishing the results of type (ii):
(ii) G is a small graph and H is a graph from some infinite family. In [8], the crossing
numbers of G × Cn for any graph G of order four except S3 were studied by Beineke and
Ringeisen, this gap was bridged by Jendrol’ et al. in [9]. The crossing numbers of Cartesian
products of 4-vertex graphs with Pn and Sn are determined by Kles̆c̆ in [10], he also determined
the crossing numbers of G × Pn for any graph G of order five [11-13]. For several special graphs
of order five, the crossing numbers of their products with Cn or Sn are also known, most of
which are due to Kles̆c̆ [14-17]. For special graphs G of order six, Peng et al. determined the
crossing number of the Cartesian product of the Petersen graph P (3, 1) with Pn in [18], Zheng
et al. gave the bound for the crossing number of Km ×Pn for m > 3, n > 1, and they determined
the exact value for cr(K6 × Pn ), see [19], and the authors [20] established the crossing number
of the Cartesian product of Pn with the complete bipartite graph K2,4 .
(iii) Both G and H belong to some infinite family. One very long attention-getting problem
of this type is to determine the crossing number of the Cartesian product of two cycles, Cm
and Cn , which was put forward by Harary et al. [3], and they conjectured that cr(Cm × Cn ) =
(m − 2)n for n ≥ m. In the next three decades, many authors were devoted to this problem and
the conjecture has been proved true for m = 3, 4, 5, 6, 7, see [8,21-24]. In 2004, the problem was
progressed by Glebsky and Salazar, who proved that the crossing number of Cm × Cn equals
its long-conjectured value for n ≥ m(m + 1) [25]. Besides the Cartesian product of two cycles,
there are several other results. D.Bokal [26] determined the crossing number of the Cartesian
product Sm × Pn for any m ≥ 3 and n ≥ 1 used a quite newly introduced operation: the zip
product. Tang, et al. [27] and Zheng, et al. [28] independently proved that the crossing number
of K2,m × Pn is 2n⌊ m m−1
2 ⌋⌊ 2 ⌋ for arbitrary m ≥ 2 and n ≥ 1.
Stimulated by these results, we begin to investigate the crossing number of the Cartesian
product of star Sn with a 6-vertex graph G2 shown in Figure 1, and get its crossing number is
6⌊ n2 ⌋⌊ n−1 n
2 ⌋ + 2n + 2⌊ 2 ⌋, for n ≥ 1.
 The Crossing Number of the Cartesian Product of Star Sn with a 6-Vertex Graph 89

v3 v4

v2 v5

v1 v6
Figure1: The graph G2 Figure 2: The graph Hn

§2. Some Basic Lemmas and the Main Result

Let A and B be two disjoint subsets of E. In a drawing D, the number of crossings made by an
edge in A and another edge in B is denoted by crD (A, B). The number of crossings made by
two edges in A is denoted by crD (A). So cr(D) = crD (E). By counting the number of crossings
in D, we have Lemma 1.

Lemma 1 Let A, B, C be mutually disjoint subsets of E. Then

crD (A ∪ B, C) = crD (A, C) + crD (B, C);

(1)
crD (A ∪ B) = crD (A) + crD (B) + crD (A, B).

The crossing numbers of the complete bipartite graph Km,n were determined by Kleitman [29]
for the case m ≤ 6. More precisely, he proved that

cr(Km,n ) = ⌊ m m−1 n n−1

2 ⌋⌊ 2 ⌋⌊ 2 ⌋⌊ 2 ⌋, if m ≤ 6 (2)

For convenience, ⌊ m m−1 n n−1

2 ⌋⌊ 2 ⌋⌊ 2 ⌋⌊ 2 ⌋ is often denoted by Z(m, n) in our paper. To ob-
tain the main result of the paper, first we construct a graph Hn which is shown in Figure
2. Let V (Hn ) = {v1 , v2 , v3 , v4 , v5 , v6 ; t1 , t2 , · · · , tn }, E(Hn ) = {vi tj | 1 6 i 6 6; 1 6 j 6
n} ∪ {v1 v2 , v1 v3 , v1 v5 , v1 v6 , v2 v3 , v3 v4 , v3 v5 , v4 v5 , v4 v6 , v5 v6 }. Let T i be the subgraph of Hn in-
duced by the edge set {vi tj |1 6 j 6 n}, and let ti be the vertex of T i of degree six. Clearly,
the induced subgraph [v1 , v2 , · · · , v6 ] ∼ = G2 . Thus, we have
n
[
Hn = G2 ∪ K6,n = G2 ∪ ( T i) (3)
i=1

For a graph G, the removal number r(G) of G is the smallest nonnegative integer r such
that the removal of some r edges from G results in a planar subgraph of G. By removing an
edge from each crossing of a drawing of G in the plane we get a set of edges whose removal
leaves a planar graph. Thus we have the following.

Lemma 2 For any drawing D of G, cr(D) > r(G).

Lemma 3 cr(H1 ) = 1, cr(H2 ) = 4.

90 Jing Wang and Yuanqiu Huang

Proof A good drawing of H1 in Figure 3 shows that cr(H1 ) 6 1, and a good drawing of
H2 in Figure 4 shows that cr(H2 ) 6 4. By Lemma 2, we only need to prove that r(H1 ) > 1 and
r(H2 ) > 4.

Figure 3: A good drawing of H1 Figure 4: A good drawing of H2

Let r = r(H1 ) and let H1′ be a planar subgraph of H1 having 16 − r edges. It is easy to see
that H1′ is a connected spanning subgraph of H1 . By Euler’s formula, in any planar drawing of
H1′ , there are 11 − r faces. Since H1′ has girth at least 3, 2(16 − r) > 3(11 − r), so r > 1, that
is r(H1 ) > 1. Similarly, we can have r(H2 ) > 4. 
In a drawing D, if an edge is not crossed by any other edge, we say that it is clean in D;
if it is crossed by at least one edge, we say that it is crossed in D.

Lemma 4 Let D be a good drawing of Hn . If there are two different subgraphs T i and T j such
that crD (T i , T j ) = 0, then crD (G2 , T i ∪ T j ) > 4.

Proof We label the vertices of G2 , see Figure 1. Since the two subgraphs T i and T j do
not cross each other in D, the induced drawing D|T i ∪T j of T i ∪ T j divides the plane into six
regions that there are exactly two vertices of G2 on the boundary of each region.
Assume to the contrary that crD (G2 , T i ∪ T j ) 6 3. The degrees of vertices v1 , v3 and
v5 in G2 are all 4, so there are at least two crossings on the edges incident to v1 , v3 and v5 ,
respectively. We can assert that edges v1 v3 , v3 v5 and v1 v5 must be crossed. Otherwise, without
loss of generality, we may assume that the edge v1 v3 is clean, then the vertices v1 and v3 must
lie on the boundary of the same region, and there are at least two crossings on the edges (except
the edge v1 v3 ) incident to vertices v1 and v3 , respectively, a contradiction. Since the degree of
vertex v4 in G2 is 3, one can easily see that there is at least one more crossing on the edges
incident to v4 , contradicts to our assumption and completes the proof. 
To obtain our main result, the following theorem is introduced.

Theorem 1 cr(Hn ) = Z(6, n) + n + 2⌊ n2 ⌋, for n > 1.

Proof A good drawing in Figure 2 shows that cr(Hn ) ≤ Z(6, n) + n + 2⌊ n2 ⌋. Now we prove
the reverse inequality by induction on n. By Lemma 3, the cases hold for n = 1 and n = 2.
Now suppose that n > 3, and for all l < n, there is

cr(Hl ) ≥ Z(6, l) + l + 2⌊ 2l ⌋ (4)

 The Crossing Number of the Cartesian Product of Star Sn with a 6-Vertex Graph 91

and for a certain good drawing D of Hn , assume that

crD (Hn ) < Z(6, n) + n + 2⌊ n2 ⌋ (5)

The following two cases are discussed:

Case 1. Suppose that there are at least two different subgraphs T i and T j that do not cross
each other in D. Without loss of generality, assume that crD (T n−1 , T n ) = 0. By Lemma 4,
crD (G2 , T n−1 ∪ T n ) ≥ 4. As cr(K3,6 ) = 6, for all i, i = 1, 2, · · · , n − 2, crD (T i , T n−1 ∪ T n ) > 6.
Using (1), (2), (3) and (4), we have
n−2
[
crD (Hn ) = crD (G2 ∪ T i ∪ T n−1 ∪ T n )
i=1
n−2
[
= crD (G2 ∪ T i ) + crD (T n−1 ∪ T n ) + crD (G2 , T n−1 ∪ T n )
i=1
n−2
X
+ crD (T i , T n−1 ∪ T n )
i=1
n−2
> Z(6, n − 2) + (n − 2) + 2⌊ ⌋ + 4 + 6(n − 2)
2
n
= Z(6, n) + n + 2⌊ ⌋
2
This contradicts (5).
Case 2. Suppose that crD (T i , T j ) > 1 for any two different subgraphs T i and T j , 1 6 i 6=
j 6 n. Using (1), (2) and (3), we have
n
[ n
[
crD (Hn ) = crD (G2 ) + crD ( T i ) + crD (G2 , T i)
i=1 i=1 (6)
n
X
> crD (G2 ) + Z(6, n) + crD (G2 , T i )
i=1

This, together with (5) implies that

n
X n
crD (G2 ) + crD (G2 , T i ) < n + 2⌊ ⌋
i=1
2

So, there is at least one subgraph T i that crD (G2 , T i ) 6 1.

Subcase 2.1 Suppose that there is at least one subgraph T i that do not cross the edges of G2 .
Without loss of generality, we may assume that crD (G2 , T n ) = 0. Let us consider the 6-cycle
C6 of the graph G2 . Hence G2 consists of C6 and four additional edges.

Subcase 2.1.1 Suppose that the edges of C6 do not cross each other in D. Since crD (G2 , T n ) =
0, then the possibility of C6 ∪ T n must be as shown in Figure 5(1). Consider the four additional
edges of G2 , they cannot cross the edges of T n and the edges of C6 either, so the unique
possibility is crD (G2 ∪ T n ) = 2, see Figure 5(1). Consider now a subdrawing of G2 ∪ T n ∪ T i of
the drawing D for some i ∈ {1, 2, . . . , n − 1}. If ti locates in the region labeled ω, then we have
92 Jing Wang and Yuanqiu Huang

crD (G2 , T i ) > 4, using crD (T n , T i ) > 1, we get crD (G2 ∪ T n , T i ) > 5. If ti locates in the other
regions, one can see that on the boundary of these regions there are at most three vertices of
G2 , and there are at least two vertices of G2 are in a region having no common edge with it, in
this case we have crD (G2 ∪ T n , T i ) > 5. Using (1), (2) and (3), we can get
n−1
[
n
crD (Hn ) = crD (G2 ∪ T ∪ T i)
i=1
n−1
[ n−1
X
= crD (G2 ∪ T n ) + crD ( T i) + crD (G2 ∪ T n , T i )
i=1 i=1
> 1 + Z(6, n − 1) + 5(n − 1)
n
> Z(6, n) + n + 2⌊ ⌋
2
which contradicts (5).

a a a
b b e
c c f
tn tn tn tn
ω d f b
f e c
e d d

(1) (2) (3) (4)

Figure 5

Subcase 2.1.2 Suppose that the edges of C6 cross each other in D. By the above arguments
in Subcase 2.1.1, we can assert that in D there must exist a subgraph T i , i ∈ {1, 2, · · · , n − 1},
such that crD (G2 ∪ T n , T i ) 6 4. The condition crD (G2 , T n ) = 0 implies that crD (C6 , T n ) = 0.
In this case the vertex tn of T n lies in the region with all six vertices of C6 on its boundary, and
the condition crD (G2 ∪ T n , T i ) 6 4 enforces that in the subdrawing of C6 ∪ T n there is a region
with at least three vertices of C6 on its boundary. In this case C6 cannot have more than two
internal crossings. If C6 has only one internal crossing, then the possibilities of C6 ∪ T n are
shown in Figure 5(2) and Figure 5(3). If C6 has two internal crossings, then the possibility of
C6 ∪ T n is shown in Figure 5(4). The vertices of G2 are labeled by a, b, c, d, e, f, respectively.
Since crD (G2 , T n ) = 0, the four edges of G2 not in C6 do not cross the edges of T n .
Consider the case shown in Figure 5(2). The three possible edges ac, ce, ae and the fourth
possible edge bd or bf or df separate the subdrawing of G2 ∪ T n into several regions with at
most three vertices of G2 on each boundary. The three possible edges bd, bf, df and the fourth
possible edge ac or ce or ae separate the subdrawing of G2 ∪ T n into several regions with at
most three vertices of G2 on each boundary. If the vertex ti of T i locates in the region with
three vertices of G2 on its boundary, one can note that there are at least 2 vertices of G2 do not
on the boundary of its neighborhood region, then crD (G2 ∪ T n , T i ) > 5; if the vertex ti of T i
locates in the region with at most two vertices of G2 on its boundary, one can see that there is at
 The Crossing Number of the Cartesian Product of Star Sn with a 6-Vertex Graph 93

least one vertex of G2 is in a region having no common edge with it, then crD (G2 ∪ T n , T i ) > 5,
a contradiction. If the possibility of C6 ∪ T n is as shown in Figure 5(3) or Figure 5(4), then a
similar contradiction can be made by the analogous arguments.

Subcase 2.2 Suppose that crD (G2 , T i ) > 1 for 1 6 i 6 n. Together with our former assump-
tion, there is at least one subgraph T i that crD (G2 , T i ) = 1. Without loss of generality, assume
that crD (G2 , T n ) = 1.

Subcase 2.2.1 Suppose that crD (C6 , T n ) = 0. Then the possibilities of C6 ∪ T n are shown
in Figure 5. It is clear that, in each region whose boundary composed of segments of edges
that incident with tn , there are at most two vertices of G2 . Adding the four additional possible
edges of G2 that have one crossing with the edges of T n , then there are at most three vertices
of G2 on the boundary of each region. Consider now a subdrawing of G2 ∪ T n ∪ T i of the
drawing D for some i ∈ {1, 2, . . . , n − 1}. If ti locates in one of the regions with three vertices
of G2 on its boundary, then then we have crD (G2 , T i ) > 3, using crD (T n , T i ) > 1, we have
crD (G2 ∪ T n , T i ) > 4. If ti locates in one of the regions with at most two vertices of G2 on its
boundary, then one can see that there are at least two vertices of G2 are in a region having no
common edge with it, in this case we have crD (G2 ∪ T n , T i ) > 6. Let

M = {T i |ti lies in the region with three vertices of G2 on its boundary}

Using (1), (2) and (3), we have

n−1
[
crD (Hn ) = crD (G2 ∪ T n ∪ T i)
i=1
n−1
[ X
= crD (G2 ∪ T n ) + crD ( T i) + crD (G2 ∪ T n , T i )
i=1 T i ∈M
X
n i
+ crD (G2 ∪ T , T )
T i ∈M
/
> 1 + Z(6, n − 1) + 4|M | + 6(n − 1 − |M |)

Together with (5), we can get

n n−1 n
2|M | > 5n − 5 − 2⌊ ⌋ − 6⌊ ⌋ > 2⌊ ⌋ (7)
2 2 2
Combined with (6) and (7), we can get
n
[ n
[
crD (Hn ) = crD (G2 ) + crD ( T i ) + crD (G2 , T i)
i=1 i=1
[n X X
= crD (G2 ) + crD ( T i) + crD (G2 , T i ) + crD (G2 , T i )
i=1 T i ∈M T i ∈M
/
> Z(6, n) + 3|M | + (n − |M |)
n
> Z(6, n) + n + 2⌊ ⌋
2
which contradicts (5).
94 Jing Wang and Yuanqiu Huang

Subcase 2.2.2 Suppose that crD (C6 , T n ) = 1, then the subdrawing of C6 ∪ T n must be one
of the ten possibilities shown in Figure 6. Adding the four additional possible edges of G2
that do not cross T n , it is not difficult to see that there are at most three vertices of G2 on
the boundary of every region. Consider now a subdrawing of G2 ∪ T n ∪ T i of the drawing D
for some i ∈ {1, 2, . . . , n − 1}. One can see that the number of crossings between the edges of
G2 ∪ T n and the edges of T i are divided into two classes:

(1) In the subdrawing of G2 ∪ T n , we have crD (G2 ∪ T n , T i ) > 5 no matter which region
does ti locate in, then a contradiction can be made by the similarly arguments in Subcase 2.1.1.
(2) In the subdrawing of G2 ∪ T n , crD (G2 ∪ T n , T i ) = 4 when ti locates in the region with
three vertices of G2 on its boundary (and crD (G2 ∪ T n , T i ) = 4 if and only if crD (G2 , T i ) = 3
and crD (T n , T i ) = 1), and crD (G2 ∪ T n , T i ) > 6 when ti locates in the other regions, then a
contradiction can be made by the similarly arguments in Subcase 2.2.1. That completes the
proof of the theorem. 

(1) (2) (3) (4) (5)

(6) (7) (8) (9) (10)

n
Figure 6: Ten possibilities of C6 ∪ T

Lemmas 5 and 6 are trivial observations.

Lemma 5 If there exists a crossed edge e in a drawing D and deleting it results in a new
drawing D∗ , then cr(D) > cr(D∗ ) + 1.

Lemma 6 If there exists a clean edge e = uv in a drawing D and contracting it into a vertex
 The Crossing Number of the Cartesian Product of Star Sn with a 6-Vertex Graph 95

u = v results in a new drawing D∗ , then cr(D) > cr(D∗ ).

Let H be a graph isomorphic to G2 . Consider a graph GH obtained by joining all vertices

of H to six vertices of a connected graph G such that every vertex of H will only be adjacent
to exactly one vertex of G. Let G∗H be the graph obtained from GH by contracting the edges
of H.
x2
x3
x1 x9
x7 x10
x6 x8
x4
x5
Figure 7

x2 x2
x3
x1 x9
x7 x10 ✲ 2x7
x8
x6 x8 x4 x6
x5 h x5

Figure 8

x2
x3 x3
x1 x9 x1
2x9
x7 x10 ✲ h
x8
x6 x8 x4 x4
x5

Figure 9

Lemma 7 cr(G∗H ) 6 cr(GH ) − 1.

Proof Let D be an optimal drawing of GH . The subgraph H has ten edges and let
x1 , x2 , . . . , x9 , x10 denote the numbers of crossings on the edges of H, see Figure 7. The following
two cases are distinguished.

Case 1. Suppose that at least one of x1 , x2 , . . . , x6 , x10 is greater than 0, then either x7 <
x1 + x3 + x4 + x9 + x10 or x9 < x2 + x5 + x6 + x7 + x10 holds. Figure 8 shows that H can be
contracted to the vertex h with at least one crossing decreased if x7 < x1 + x3 + x4 + x9 + x10 .
Figure 9 shows that H can be contracted to the vertex h with at least one crossing decreased
if x9 < x2 + x5 + x6 + x7 + x10 . That means cr(G∗H ) 6 crD (GH ) − 1 = cr(GH ) − 1.
96 Jing Wang and Yuanqiu Huang

x2 h
x1 x9 x3 2x1 x3
x7 x10 ✲
x10
x6 x8 x4 x6
x5

Figure 10

Case 2. Suppose that x1 = x2 = · · · = x6 = x10 = 0, then we have x7 + x8 + x9 > 1 since

cr(H1 ) = 1. Figure 10 shows that H can be contracted to the vertex h in the following way:
first, delete the edges x7 , x8 and x9 , (for convenience, here we use xi to denote the respective
edge with xi crossings), then redraw the former edge x7 closely enough to edges x1 and x6 , at
last, contract the edge x2 into a vertex h. By Lemma 5, the first step decreases at least one
crossing. And by Lemma 6, the second and last steps do not increase the number of crossings.
That means cr(G∗H ) 6 crD (GH ) − 1 = cr(GH ) − 1. This completes the proof. 

Consider now the graph G2 × Sn . For n > 1 it has 6(n + 1) vertices and edges that are the
edges in n + 1 copies Gi2 , i = 0, 1, · · · , n, and in the six stars Sn , see Figure 11.

Figure 11: A good drawing of G2 × Sn

Now, we can get the main theorem.

Theorem 2 cr(G2 × Sn ) = Z(6, n) + 2n + 2⌊ n2 ⌋, for n > 1.

Proof A drawing in Figure 11 shows that cr(G2 × Sn ) 6 Z(6, n) + 2n + 2⌊ n2 ⌋. Assume

that there is an optimal drawing D of G2 × Sn with fewer than Z(6, n) + 2n + 2⌊ n2 ⌋ crossings.
Contracting the edges of each Gi2 to a vertex ti for all i = 1, 2, · · · , n in D results in a graph
homeomorphic to Hn , and using Lemma 7 repeatedly, we have cr(Hn ) ≤ cr(G2 × Sn ) − n =
 The Crossing Number of the Cartesian Product of Star Sn with a 6-Vertex Graph 97

crD (G2 ×Sn )−n < Z(6, n)+n+2⌊ n2 ⌋, a contradiction with Theorem 1. Therefore, cr(G2 ×Sn ) =
Z(6, n) + 2n + 2⌊ n2 ⌋. 

References

[1] F. Harary, Graph Theory (Addision-Wesley, Reading, MA, 1969).

[2] M. R. Garary and D. S. Johnson, Crossing number is NP-complete, SIAM J. Algebric
Discrete Methods 4, (1993) 312-316.
[3] F.Harary, P.C.Kainen, A.J.Schwenk, Toroidal graphs with arbitrarily high crossing num-
bers, Nana Math. 6, (1973) 58-67.
[4] A. M. Dean, R. B. Richter, The crossing number of C4 × C4 , J. Graph Theory 19, (1995)
125-129.
[5] Richter, B.R., Thomassen, C., Intersection of curve systems and the crossing number of
C5 × C5 , Discrete and Computational Geometry 13, (1995) 149-159.
[6] Anderson, M.S., Richter, B.R., Rodney, P., Crossing number of C6 × C6 , Congressus Nu-
merantium 118, (1996) 97-107.
[7] M. Kles̆c̆, The crossing number of K2,3 × C3 , Discrete Math. 251, (2002) 109-117.
[8] L. W. Beineke, R. D. Ringeisen, On the crossing numbers of products of cycles and graphs
of order four, J. Graph Theory 4, (1980) 145-155.
[9] S. Jendrol’, M.S̆c̆erbová, On the crossing numbers of Sm × Pn and Sm × Cn , C̆as. Pest.
Mat. 107, (1982) 225-230.
[10] M. Kles̆c̆, The crossing numbers of products of paths and stars with 4-vertex graphs, J.
Graph Theory 18, (1994) 605-614.
[11] M. Kles̆c̆, The crossing numbers of certain Cartesian products, Discuss. Math. - Graph
Theory 15, (1995) 5-10.
[12] M. Kles̆c̆, The crossing number of K5 × Pn , Tatra Moutains Math. Publ. 18, (1999) 63-68.
[13] M. Kles̆c̆, The crossing number of Cartesian products of paths with 5-vertex graphs, Dis-
crete Math. 233, (2001) 353-359.
[14] M. Kles̆c̆, The crossing number of K2,3 × Pn and K2,3 × Sn , Tatra Moutains Math. Publ.
9, (1996) 51-56.
[15] M. Kles̆c̆, On the crossing numbers of Cartesian products of stars and paths or cycles,
Math. Slovaca 41, (1991) 113-120.
[16] M.Kles̆c̆, On the crossing numbers of products of stars and graphs of order five, Graphs
and Combinatorics 17, (2001) 289-294.
[17] M. Kles̆c̆, A. Kocúrová, The crossing numbers of products of 5-vertex graphs with cycles,
Discrete Math. 307, (2007) 1395-1403.
[18] Y. H. Peng, Y. C. Yiew, The crossing number of P (3, 1) × Pn , Discrete Math. 306, (2006)
1941-1946.
[19] Zheng Wenping, Lin Xiaohui, Yang Yuansheng, Cui Chong, On the crossing number of
Km × Pn , Graphs and Combinatorics 23, (2007) 327-336.
[20] Wang Jing, Huang Yuanqiu, The Crossing Number of K2,4 ×Pn , Acta Mathematica Scientia
(in Chinese) 28A, (2008) 251-255.
98 Jing Wang and Yuanqiu Huang

[21] R. D. Ringeisen, L. W. Beineke, The crossing number of C3 × Cn , J. Combin. Theory Ser.

B 24, (1978) 134-136.
[22] M. Kles̆c̆, R. B. Richter, I. Stobert, The crossing number of C5 × Cn , J. Graph Theory 22,
(1996) 239-243.
[23] R. B. Richter, G. Salazar, The crossing number of C6 × Cn , Australas. J. Combin. 23,
(2001) 135-143.
[24] J. Adamsson, R. B. Richter, Arrangements, circular arrangements, and the crossing number
of C7 × Cn , J. Combin. Theory Ser. B 90, (2004) 21-39.
[25] L.Y.Glebsky, G.Salazar, The crossing number of Cm × Cn is as conjectured for n ≥ m(m +
1), J.Graph Theory 47, (2004) 53-72.
[26] D. Bokal, On the crossing numbers of Cartesian products with paths, J. Combin. Theory
Ser B. 97, (2007) 381-384.
[27] Tang Ling, Lv Shengxiang, Huang Yuanqiu, The crossing number of Cartesian products
of complete bipartite graphs K2,m with paths Pn , Graphs and Combinatorics 23, (2007)
659-666.
[28] Zheng Wenping, Lin Xiaohui, Yang Yuansheng, The crossing number of K2,m ×Pn , Discrete
Math. 308, (2008) 6639-6644.
[29] D.J Kleitman, The crossing number of K5,n , J.Comb.Theory 9, (1970) 315-323.
Math.Combin.Book Ser. Vol.1 (2010), 99-102

Smarandache’s Pedal Polygon Theorem

in the Poincaré Disc Model of Hyperbolic Geometry

Cătălin Barbu

(”Vasile Alecsandri” High School - Romania)

Email: kafka [email protected]

Abstract: In this note, we present a proof of the hyperbolic a Smarandache’s pedal

polygon theorem in the Poincaré disc model of hyperbolic geometry.

Keywords: Hyperbolic geometry, hyperbolic triangle, gyro-vector, hyperbolic Pytagorean

theorem.

AMS(2010): 30F45, 20N99, 51B10, 51M10

§1. Introduction

Hyperbolic Geometry appeared in the first half of the 19th century as an attempt to understand
Euclid’s axiomatic basis of Geometry. It is also known as a type of non-Euclidean Geometry,
being in many respects similar to Euclidean Geometry. Hyperbolic Geometry includes similar
concepts as distance and angle. Both these geometries have many results in common but many
are different.
There are known many models for Hyperbolic Geometry, such as: Poincaré disc model,
Poincaré half-plane, Klein model, Einstein relativistic velocity model, etc. In this note we choose
the Poincaré disc model in order to present the hyperbolic version of the Smarandache’s pedal
polygon theorem. The Euclidean version of this well-known theorem states that if the points
Mi , i = 1, n are the projections of a point M on the sides Ai Ai+1 , i = 1, n, where An+1 = A1 , of
the polygon A1 A2 ...An , then M1 A21 +M2 A22 +...+MnA2n = M1 A22 +M2 A23 +...+Mn−1 A2n +Mn A21
[1]. This result has a simple statement but it is of great interest.
We begin with the recall of some basic geometric notions and properties in the Poincaré
disc. Let D denote the unit disc in the complex z - plane, i.e.

D = {z ∈ C : |z| < 1}

The most general Möbius transformation of D is

z0 + z
z → eiθ = eiθ (z0 ⊕ z),
1 + z0 z
1 Received February 20, 2010. Accepted March 28, 2010.
100 Cătălin Barbu

which induces the Möbius addition ⊕ in D, allowing the Möbius transformation of the disc to
be viewed as a Möbius left gyro-translation
z0 + z
z → z0 ⊕ z =
1 + z0 z
followed by a rotation. Here θ ∈ R is a real number, z, z0 ∈ D, and z0 is the complex conjugate
of z0 . Let Aut(D, ⊕) be the automorphism group of the groupoid (D, ⊕). If we define
a⊕b 1 + ab
gyr : D × D → Aut(D, ⊕), gyr[a, b] = = ,
b⊕a 1 + ab
then is true gyro-commutative law

a ⊕ b = gyr[a, b](b ⊕ a).

A gyro-vector space (G, ⊕, ⊗) is a gyro-commutative gyro-group (G, ⊕) that obeys the

following axioms:
(1) gyr[u, v]a· gyr[u, v]b = a · b for all points a, b, u, v ∈G.
(2) G admits a scalar multiplication, ⊗, possessing the following properties. For all real
numbers r, r1 , r2 ∈ R and all points a ∈G:

(G1) 1 ⊗ a = a;
(G2) (r1 + r2 ) ⊗ a = r1 ⊗ a ⊕ r2 ⊗ a;
(G3) (r1 r2 ) ⊗ a = r1 ⊗ (r2 ⊗ a);
|r|⊗a a
(G4) kr⊗ak = kak ;
(G5) gyr[u, v](r ⊗ a) = r ⊗ gyr[u, v]a;
(G6) gyr[r1 ⊗ v, r1 ⊗ v] =1 ;
(3) Real vector space structure (kGk , ⊕, ⊗) for the set kGk of one-dimensional ”vectors”

kGk = {± kak : a ∈ G} ⊂ R

with vector addition ⊕ and scalar multiplication ⊗, such that for all r ∈ R and a, b ∈ G,
(G7) kr ⊗ ak = |r| ⊗ kak;
(G8) ka ⊕ bk ≤ kak ⊕ kbk.

Definition 1.1 The hyperbolic distance function in D is defined by the equation

a−b
d(a, b) = |a ⊖ b| = .
1 − ab
Here, a ⊖ b = a ⊕ (−b), for a, b ∈ D.

Theorem 1.2(The Möbius Hyperbolic Pythagorean Theorem) Let ABC be a gyrotriangle in

a Möbius gyrovector space (Vs , ⊕, ⊗), with vertices A, B, C ∈ Vs , sides a, b, c ∈ Vs and side
gyrolenghts a, b, c ∈ (−s, s), a = −B ⊕ C, b = −C ⊕ A, c = −A ⊕ B, a = kak , b = kbk , c = kck
and with gyroangles α, β, and γ at the vertices A, B, and C. If γ = π/2, then
c2 a2 b2
= ⊕
s s s
 Smarandache’s Pedal Polygon Theorem in the Poincaré Disc Model of Hyperbolic Geometry 101

(see [2, p 290])

For further details we refer to the recent book of A.Ungar [2].

§2. Main Result

In this sections, we present a proof of the hyperbolic a Smarandache’s pedal polygon theorem
in the Poincaré disc model of hyperbolic geometry.

Theorem 2.1 Let A1 A2 ...An be a hyperbolic convex polygon in the Poincaré disc, whose
vertices are the points A1 , A2 , ..., An of the disc and whose sides (directed counterclockwise) are
a1 = −A1 ⊕ A2 , a2 = −A2 ⊕ A3 , ..., an = −An ⊕ A1 . Let the points Mi , i = 1, n be located on the
sides a1 , a2 , ..., an of the hyperbolic convex polygon A1 A2 ...An respectively. If the perpendiculars
to the sides of the hyperbolic polygon at the points M1 , M2, ..., and Mn are concurrent, then the
following equality holds:
2 2 2 2 2 2
|−A1 ⊕ M1 | ⊖|−M1 ⊕ A2 | ⊕|−A2 ⊕ M2 | ⊖|−M2 ⊕ A3 | ⊕...⊕|−An ⊕ Mn | ⊖|−Mn ⊕ A1 | = 0.

Proof We assume that perpendiculars meet at a point of A1 A2 ...An and let denote this
point by M (see Figure). The geodesic segments −A1 ⊕ M, −A2 ⊕ M, ..., −An ⊕ M, −M1 ⊕ M,
−M2 ⊕ M, ..., −Mn ⊕ M split the hyperbolic polygon into 2n right-angled hyperbolic triangles.
We apply the Theorem 1.2 to these 2n right-angled hyperbolic triangles one by one, and we
easily obtain:
2 2 2
|−M ⊕ Ak | = |−Ak ⊕ Mk | ⊕ |−Mk ⊕ M | ,
for all k from 1 to n, and M0 = Mn .Using equalities
2 2
|−M ⊕ Ak | = |−Ak ⊕ M | , k = 1, n,

we have
2 2 2 2 ′
αk = |−Ak ⊕ Mk | ⊕ |−Mk ⊕ M | = |−M ⊕ Mk−1 | ⊕ |−Mk−1 ⊕ Ak | = αk
102 Cătălin Barbu

for all k from 1 to n, and M0 = Mn . This implies

′ ′ ′
α1 ⊕ α2 ⊕ ... ⊕ αn = α1 ⊕ α2 ⊕ ... ⊕ αn .

Since ((−1, 1), ⊕) is a commutative group, we immediately obtain

2 2 2 2 2 2
|−A1 ⊕ M1 | ⊕|−A2 ⊕ M2 | ⊕...⊕|−An ⊕ Mn | = |−M1 ⊕ A2 | ⊕|−M2 ⊕ A3 | ⊕...⊕|−Mn ⊕ A1 | ,

i.e.
2 2 2 2 2 2
|−A1 ⊕ M1 | ⊖|−M1 ⊕ A2 | ⊕|−A2 ⊕ M2 | ⊖|−M2 ⊕ A3 | ⊕...⊕|−An ⊕ Mn | ⊖|−Mn ⊕ A1 | = 0.

References

[1] F. Smarandache, Problémes avec et sans. . . probléms!, pp. 49 & 54-60, Somipress, Fés,
Morocoo, 1983.
[2] Ungar, A.A.,Analytic Hyperbolic Geometry and Albert Einstein’s Special Theory of Rela-
tivity, Hackensack, NJ:World Scientific Publishing Co.Pte. Ltd., 2008.
Math.Combin.Book Ser. Vol.1 (2010), 103-114

The Arc Energy of Digraph

Z.Khoshbakht

(Department of Studies in Mathematics, University of Mysore, Manasagangothri, Mysore - 570 006, INDIA)

Email: [email protected]

Abstract: We study the energy of the arc-adjacency matrix of a directed graph D, which
is simply called the arc energy of D. In particular, we give upper and lower bounds for the
arc energy of D. We show that arc energy of a directed tree is independent of its orientation.
We also compute arc energies of directed cycles and some unitary cayley digraphs.

Keywords: Smarandache arc k-energy, digraph, arc adjacency matrix, arc energy.

AMS(2010): 05C50

§1. Introduction

Let D be a simple digraph with vertex set V (D) = {v1 , v2 , · · · , vn } and arc set Γ(D) ⊂ V (D) ×
V (D). Let |Γ(D)| = m. The arc adjacency matrix of D is the n × n matrix A = [aij ], where


1 if i < j and (vi , vj ) ∈ Γ(D)

aij = −1 if i < j and (vj , vi ) ∈ Γ(D)



0 if vi and vj are not adjacent.

For i > j we define aij = aji . A is a symmetric matrix of order n and all its eigenvalues are
real. We denote the eigenvalues of A by λ1 , λ2 , · · · , λn with λ1 ≥ λ2 ≥ · · · ≥ λn . The set
{λ1 , λ2 , · · · , λn } is called the arc spectrum of D. The characteristic polynomial |xI − A| of the
arc adjacency matrix A is called the arc characteristic polynomial of D and it is denoted by
Φ(D; x). The arc energy of D is defined by
n
X
Ea (D) = |λi |.
i=1

For the majority of conjugated hydrocarbons, The total π−electron energy, Eπ satisfies
the relation
n
X
Eπ (D) = |λi |
i=1

1 Received February 28, 2010. Accepted March 30, 2010.

104 Z.Khoshbakht

where λ1 , λ2 , · · · , λn are the eigenvalues of the molecular graph of the conjugated hydrocarbons.
In view of this, Gutman [3] introduced the concept of graph energy E(G) of a simple undirected
graph G and he defined it as
Xn
E(G) = |λi |
i=1

where λ1 , λ2 , · · · , λn are the eigenvalues of the adjacency matrix of G. Survey of development

of this topic before 2001 can be found in [4]. For recent development, one can consult [2]. The
energy of a graph has close links to chemistry [5]. In many situations chemists use digraph
rather than graphs. In this paper we are interested in studying mathematical aspects of arc
energy of digraphs. The skew energy of a digraph is recently studied in [1].
In Section 2 of this paper we study some basic properties of the arc energy and also derive
an upper bound for Ea (D). In Section 3 we study arc energy of directed trees. We compute arc
energies of directed cycles and some unitary Cayley digraphs in Section 4 and 5 respectively.

§2. Basic Properties of Arc Energy

We begin with the definition of arc energy.

Definition 2.1 Let A be the arc adjacency matrix of a digraph D. Then its Smarandache arc
Pn
k-energy EaK (D) is defined as i=1 |λi |k , where n is the order of D and λi , 1 ≤ i ≤ n are the
eigenvalues of A. Particularly, if k = 1, the Smarandache arc k-energy Ea1 (D) is called the arc
energy of D and denoted by Ea (D) for abbreviatation.

Example 2.2 Let D be a directed cycle on four vertices.

-
6 ?


 
0 1 0 −1
 
1 0 1 0
 
Then A =   and the characteristic polynomial of A is λ4 − 4λ2 + 4, and
0 1 0 1
 
−1 0 1 0
√ √ √ √ √
hence the eigenvalues of A are − 2, 2, − 2, 2, and the arc energy of D is 4 2.

Theorem 2.3 Let D be a digraph with the arc adjacency characteristic polynomial

Φ(D; x) = b0 xn + b1 xn−1 + · · · + bn .

Then
 The Arc Energy of Digraph 105

(i) b0 = 1;

(ii) b1 = 0;

(iii) b2 = −m, the number of arcs of D;

(iv) For i < j < k, we define

(i, j) = number of triangles of the form

vk

vi
- I vj

and

(i, j, k) = number of triangles of the form

vk

vi - I vj

b3 = −2[(i, j) + (j, k) + (k, i) + (k, j, i) − (j, i) − (k, j) − (i, k) − (i, j, k)].

Proof

(i) It follows from the definition, Φ(D; x) = det(xI − A), that b0 = 1.

(ii) Since the diagonal elements of A are all zero, the sum of determinants of all 1 × 1 principal
submatrices of A = trace of A = 0. So b1 = 0.

(iii) The sum of determinants of all 2 × 2 principal submatrices of

 
X 0 ajk X X
A= det  = −ajk akj = − a2jk = −m.
j<k akj 0 j<k j<k

Thus b2 = −m.

(iv) We have

0 aij aik
3
P
b3 = (−1) i<j<k aji 0 ajk
aki akj 0
106 Z.Khoshbakht

0 aij aik
3
P
= (−1) i<j<k aij 0 ajk
aik ajk 0
P
= −2 i<j<k sij sik sjk

= −2[(i, j)+(j, k)+(k, i)+(k, j, i)−(j, i)−(k, j)−(i, k)−(i, j, k)].

Theorem 2.4 If λ1 , λ2 , · · · , λn are the arc eigenvalues of a digraph D, then

Pn
(i) i=1 λ2i = 2m;

(ii) For 1 ≤ i ≤ n, |λi | ≤ ∆, the maximum degree of the underlying graph GD .

Pn Pn Pn
Proof (i) We have i=1 λ2i = trace of A2 = i=1 j=1 aij aji
n X
X n
= (aij )2 = 2m.
i=1 j=1

(ii) Let λ1 ≥ λ2 ≥ · · · ≥ λn be the eigenvalues of A. The Cauchy-Schwartz inequality state

that if (a1 , a2 , · · · , an ) and (b1 , b2 , · · · , bn ) are real n-vectors then

n
!2 n
! n
!
X X X
ai b i ≤ a2i b2i .
i=1 i=1 i=1

Let ai = 1 and bi = |λi | for 1 ≤ i ≤ n, and i 6= j. Then

 2  
n
X Xn
   
 |λi | ≤ (n − 1)  |λi |2  . (2.1)
i=1 i=1
i6=j i6=j

Pn Pn
Since i=1 λi = 0 we have i=1, λi = −λj . Thus
i6=j

n
X
| λi |2 = | − λj |2 .
i=1,
j6=i

Hence  2
n
X
 
| − λj |2 ≤  |λi | .
i=1
i6=j

Using (2.1) in the above inequality we get

n
X

| − λj |2 ≤ (n − 1) |λi |2 − |λj |2 .
i=1
 The Arc Energy of Digraph 107

i.e.,
n|λj |2 ≤ 2m(n − 1),

|λj |2 ≤ (n − 1)2 .

Hence
|λj | ≤ ∆.

Corollary 2.5 Ea (D) ≤ n∆.

p √ √ Qn
Theorem 2.6 2m + n(n − 1)p2/n ≤ Ea (D) ≤ 2mn ≤ n ∆ where p = |det A| = i=1 |λi |.

Proof We have

n
!2 n
2
X X X
(Ea (D)) = |λi | = λ2i + |λi | |λj |
i=1 i=1 i6=j

and by the inequality between the arithmetic and geometric means,

n
! n1
1 Y 1
Ea (D) ≥ |λi | = |det A| n
n i=1

  n(n−1)
1

1 X Y
∴ |λi | |λj | ≥  |λi | |λj |
n(n − 1)
i6=j i6=j

n
! n(n−1)
1
Y
= |λi |2(n−1)
i=1

n
! n2 n
Y Y 2 2
= |λi | =| λi | n = p n .
i=1 i=1

Therefore

2 2
(Ea (D)) ≥ 2m + n(n − 1)p n .

To prove the right hand side inequality , we apply Schwartz’s inequality to the Euclidean
vectors u = (|λ1 |, |λ2 |, · · · , |λn |) and v = (1, 1, · · · , 1) to get
v
n u n
X uX √ √ √ √
Ea (D) = |λi | ≤ t |λi |2 n = 2mn ≤ n∆n = n ∆. (2.2)
i=1 i=1


√
Corollary 2.7 Ea (D) = n ∆ if and only if A2 = ∆In where In is the identity matrix of order
n.
108 Z.Khoshbakht

Proof Equality holds in (2.2) if and only if the Schwartz’s inequality becomes equality and
Pn
trace A2 = i=1 λ2i = 2m = n∆, if and only if, there exists a constant α such that |λi |2 = α
for all i and GD is a ∆-regular graph, if and only if, A2 = αIn and α = ∆. 

Theorem 2.8 Each even positive integer 2p is the arc energy of a directed star.

Proof Let V (K1,n ) = {v1 , . . . , vn+1 }. If vn+1 is the center of K1,n , orient all the edges
toward vn+1 . Then
 
0 0 ... 0 1
 
 
0 0 ... 0 1
A=  .. .. . . .

..  ,
. . . .. .
 
1 1 ... 1 0
√ √ √
and its eigenvalues are { n, − n, 0, 0, . . . , 0}, and so Ea (K1,n ) = 2 n. Now take n = p2 . 

§3. Arc Energies of Trees

We begin with a basic lemma.

Lemma 3.1 Let D be a simple digraph. and let D′ be the digraph obtained from D by reversing
the orientations of all the arcs incident with a particular vertex of D. Then Ea (D) = Ea (D′ ).

Proof Let A(D) be the arc adjacency matrix of D with respect to a labeling of its vertex
set. Suppose the orientations of all the arcs incident at vertex vi of D are reversed. Let the
resulting digraph be D′ . Then A(D′ ) = Pi A(D)Pi where Pi is the diagonal matrix obtained
from the identity matrix by changing the i-th diagonal entry to −1. Hence A(D) and A(D′ )
are orthogonally similar, and so have the same eigenvalues, and hence D and D′ have the same
arc energy. 

Lemma 3.2 Let T be a labeled directed tree rooted at vertex v. It is possible, through reversing
the orientations of all arcs incident at some vertices other than v, to transform T to a directed
tree T ′ in which the orientations of all the arcs go from low labels to high labels.

Proof The proof is by induction on n, the order of the tree. For n = 2, there is only one
arc and the result is true. Assume that any labeled directed tree of order less than n can be
transformed in the manner described to a directed tree T ′ such that the orientations of all the
arcs go from low labels to high labels. Consider a labeled directed tree T of order n rooted
at v. Let N (v) be the neighbor set of v. For each w ∈ N (v), reverse the orientations of all
the arcs incident at w, if necessary, so that the orientation of the arc between v and w is from
low to high labels. Now, by induction assumption, the old-labeled new-orientation subtree Tw
rooted at w ∈ N (v) can be transformed to a directed subtree Tw′ such that the orientations of
all the arcs go from low labels to high labels. Now combine all the subtrees Tw′ and the root v
to obtain the required tree T ′ . 
 The Arc Energy of Digraph 109

Theorem 3.3 The arc energy of a directed tree is independent of its orientation.

Proof Let T be a labeled directed tree. Since the underlying

 graph
 is a tree, it is a
0 Y
bipartite graph, and hence we can label T such that A(T ) =  . By Lemma 3.2, we
T
Y 0
 
0 X
can transform T to T ′ such that A(T ′ ) =  , where X is nonnegative. By applying
XT 0
Lemma 3.1 repeatedly, we conclude that A(T ) and A(T ′ ) are orthogonally similar, and hence
have the same eigenvalues and so the same arc energy. Consequently, T has the same arc energy
as the special directed tree T ′ in which the orientations of all the arcs go from low labels to
high labels. 

Corollary 3.4 The arc energy of a directed tree is the same as the energy of its underlying
tree.

Proof From the proof

 of Theorem
 3.3, the arc energy of a directed tree is equal to the sum
0 X
of the singular values of  , which is nothing but the adjacency matrix of underlying
T
X 0
undirected tree and so the arc energy of a directed tree is the same as the energy of its underlying
undirected tree. 

Corollary 3.5 Energy of a special tournament of order n with vertex set {1, 2, . . . , n} in which
all its arcs point from low labels to high labels is same as its underlying tournament.

§4. Computation of Arc Energies of Cycles

In this section, we compute the arc energies of cycles under different orientations. Given a
directed cycle, fix a vertex and label the vertices consecutively. Reversing the arcs incident at
a vertex if necessary, we obtain a new directed cycle with arcs going from low labels to high
labels with a possible exception of one arc. Hence the arc adjacency matrix of a directed cycle
is orthogonally similar to either A+ or A− where,

   
0 1 0 ... 1 0 1 0 ... −1
   
   
1 0 1 ... 0 1 0 1 ... 0
A+ = 
 .. .. ..
 and A− = 
..   .. .. ..

..  .
. . . .  . . . . 
   
1 0 0 ... 0 −1 0 0 ... 0

Case (i): Let Cn+ be the directed cycle with arc adjacency matrix A+ . We have A+ = Z +Z n−1
110 Z.Khoshbakht

where  
0 1 0 ... 0
 
 
0 0 1 ... 0
Z=
 .. .. ..

.. 
. . . .
 
1 0 0 ... 0
which is a circulant matrix. Since Z = I, the characteristic polynomial of Z is xn − 1. Hence
n
2πi
we have Sp(Z) = {1, w, w2 , · · · , wn−1 } where w = e n and so

Sp(Cn+ ) = {wj + wj(n−1) : j = 0, 1, 2, · · · , n − 1}

= {wj + w−j : j = 0, 1, 2, · · · , n − 1}
2jπ
= {2 cos( ) : j = 0, 1, 2, · · · , n − 1}.
n
For n = 2k + 1, we have
n−1
X Xk
2jπ 2jπ
Ea (Cn+ ) = 2| cos(
)| = 2 + 4 | cos( )|
j=0
n j=1
(2k + 1)
 
Xk
jπ sin (2k+1)π
2(2k+1) 1
= 2+4 cos( ) = 2+4 π − 
j=1
(2k + 1) 2 sin 2(2k+1) 2
π π
= 2 csc( ) = 2 csc( ).
2(2k + 1) 2n
For n = 4k,
n−1
X k−1
X
2jπ jπ
Ea (Cn+ ) = 2| cos( )| = 4 + 8 cos( )
j=0
n j=1
2k
!
sin (2k−1)π
4k 1 π π
= 4+8 π − = 4 cot( ) = 4 cot( ).
2 sin 4k 2 4k n

Similarly for n = 4k + 2
π
Ea (Cn+ ) = 4 csc( ).
n
Putting together the results above, we obtain the following formulas for arc energy of Cn+ :

 π
2 csc 2n
 if n ≡ 1(mod2),
+ π
Ea (Cn ) = 4 cot n if n ≡ 0(mod4),


 π
4 csc n if n ≡ 2(mod4).

Case (ii): Let Cn− be the directed cycle with arc adjacency matrix A− . We have A− = Z−Z n−1
where  
0 1 0 ... 0
 
 
0 0 1 . . . 0

Z = . 
.. .. ..  .
 .. . . . 
 
−1 0 0 . . . 0
 The Arc Energy of Digraph 111

Since Z n = −I, the characteristic polynomial of Z is xn + 1. Hence we have Sp(Z) =

(2j+1)πi
{e n | j = 0, 1, · · · , (n − 1)}. So Sp(A− ) = {z − z n−1 | z ∈ Sp(Z)}.

For n = 2k + 1, we have
n−1 k 2k
!
X (2j + 1)π X mπ X mπ
Ea (Cn− ) = 2| cos( )| = 2 cos( )− cos( )
j=0
2k + 1 m=0
2k + 1 2k + 1
m=k+1
k
X 2k
X  !
mπ 2k + 1 − m
= 2 1+ cos( )− cos π −
m=1
2k + 1 2k + 1
m=k+1
k
X 2k
X   !
mπ 2k + 1 − m
= 2 1+ cos( )+ cos π
m=1
2k + 1 2k + 1
m=k+1
k
! k
X mπ X mπ
= 2 1+2 cos( ) = 2+4 cos( )
m=1
2k + 1 m=1
2k +1
π
= 2 csc( ).
2n
For n = 4k, we have
n−1
X k−1
X
(2j + 1)π (2j + 1)π
Ea (Cn− ) = | cos( )| = 8 cos( )
j=0
4k j=0
4k
k
!
X (2j − 1)π sin (k+1)π
4k cos( π4 − π
4k )
= 8 cos( )=8 π .
j=1
4k sin 4k

Similarly for n = 4k + 2, we get

(k+1)π kπ π
sin( 2(2k+1) ) cos( 2(2k+1) − 2(2k+1) )
Ea (Cn− ) = π .
sin 2(2k+1)

Putting together the results above, we obtain the following formulas for arc energy of Cn− :

 π

 2 csc( 2n ) if n ≡ 1(mod2),
  sin (k+1)π cos( π − π ) 

Ea (Cn− ) = 8
4k
π
sin 4k
4 4k
if n ≡ 0(mod4),



 (k+1)π kπ π
 sin( 2(2k+1) ) cos( 2(2k+1)
π
− 2(2k+1) )
if n ≡ 2(mod4).
sin 2(2k+1)

§4. On the Arc Energies of Some Unitary Cayley Digraphs

We now define the unit Cayley digraph Dn , n > 1. The vertex set of Dn is V (Dn ) =
{0, 1, 2, · · · , (n − 1)} and the arc set of Dn is Γ(Dn ) and is defined as follows:

For i, j ∈ {0, 1, 2, · · · , (n − 1)} with i < j and (j − i, n) = 1, (i, j) ∈ Γ(Dn ) or (j, i) ∈ Γ(Dn )
according as j − i is a quadratic residue or a quadratic non-residue modulo n. In this section
we compute arc energies of unitary Cayley digraphs Dn for n = 2α0 pα αr
1 · · · pr , α0 = 0 or 1,
1
112 Z.Khoshbakht

pi ≡ 1 (mod 4), i = 1, 2, 3, · · · , r. We make use of the following well-known result to establish

a formula for arc energy of Dn for certain values of n.

Theorem 5.1 Let n = 2α0 pα αr 2

1 · · · pr , n > 1 and (a, n) = 1. Then x ≡ a (mod n) is solvable
1

if and only if
 
a
(i) pi = 1 for i = 1, 2, · · · , r

and

(ii) a ≡ 1 (mod 4) if 4 | n but 8 ∤ n ; a ≡ 1 (mod 8) if 8 | n.

 
a
Here pi is the Legendre symbol.

Theorem 5.2 For n = 2α0 pα αr

1 · · · pr , α0 = 0 or 1, pi ≡ 1 (mod 4), i = 1, 2, 3, · · · , r, the
1

arc adjacency eigenvalues of the unitary Cayley digraph Dn are the Gauss sums G(r, χf ), r =
0, 1, 2, · · · , n − 1, associated with quadratic character f .

Proof The arc adjacency matrix of Dn with respect to the natural order of the vertices
0, 1, · · · , n − 1 is
 
( n0 ) ( n1 ) ( n2 ) ... ( i−1
n ) ... ( n−1
n )

 
 
 
 (1) ( n0 ) ( n1 ) ... ( i−2 n−2 
 n n ) ... ( n )
 . 
 
An =  .. 
 
 i−1 n−i 
( n ) ( i−2
n ) ( i−3
n ) ... ( n0 ) ... ( n )
 
 .. 
 . 
 
( n−1
n ) ( n−2
n ) ( n−3
n ) ... ( n−i
n ) ... ( n0 )

where


 1 if
 (a, n) = 1 and x2 ≡ a(mod n) is solvable,
a
= −1 if (a, n) = 1 and x2 ≡ a(mod n) is not solvable,
n 


0 otherwise.

Since n = 2α0 pα αr
1 · · · pr , n > 1, where α0 = 0 or 1 and pi ≡ 1 (mod 4), i = 1, 2, 3, · · · , r, it
1

follows from Theorem 5.1 that x2 ≡ −1 (mod n) is solvable. Thus

 
1
− = 1. (5.1)
n

Moreover, if (a, n) = 1 then

        
n−a −a −1 a a
= = = ( using (5.1)).
n n n n n
 The Arc Energy of Digraph 113

Hence the arc adjacency matrix An of Dn is circulant. Consequently the eigenvalues of An are
given by
n−1
X m 2πi
λr = wrm , r = 0, 1, · · · , n − 1, w = e n

m=0
n
n−1
X m
= wrm = G(r, χf )
m=1
n

where χf is the Dirichlet quadratic character mod n. 

Theorem 5.3 If n = 2α0 pα αr

1 · · · pr , n > 1, where α0 = 0 or 1 and pi ≡ 1 (mod 4), i =
1

1, 2, · · · , r then the arc energy of Dn is

√
Ea (Dn ) = n φ(n).

Proof By Theorem 5.2, the eigenvalues of Dn are

λr = G(r, χf ), 0 ≤ r ≤ n − 1.

Hence the arc energy of Dn is given by

n−1
X n−1
X
Ea (Dn ) = |λr | = |G(r, χf )|
r=0 r=0
n−1
X
= |χf (r)||G(1, χf )| = |G(1, χf )| φ(n).
r=1

Therefore, to complete the proof, we need to compute |G(1, χf )|. We have

n
X −2πim
|G(1, χf )|2 = G(1, χf )G(1, χf ) = G(1, χf ) χf (m) e n

m=1
n
X n  
n X
X
−2πim j 2πijm −2πim
= G(m, χf ) e n = e n e n

m=1 m=1 j=1

n
Xn   n
j X 2πi
= wm(j−1) , where w = e n

j=1
n m=1
 X n Xn
1
= 1, since wm(j−1) = 0, if j > 1
n m=1 m=1
= n.
√ √
Hence |G(1, χf )| = n and E(Dn ) = n φ(n).

Conclusion The arc spectrum and arc energy of Dn when n ≡ 1 or 2(mod 4) was computed
(Theorems 5.2 and 5.3.) using fact that the associated arc adjacency matrix An was circulant.
Since in general An is not circulant, we leave open the problem of computing the arc spectrum
and arc energy of Dn for any natural number n.
114 Z.Khoshbakht

References

[1] C. Adiga, R. Balakrishnan and Wasin So, The skew energy of digraph, Linear Algebra and
its Appl. (to appear).
[2] Richard A Brualdi , Energy of a graph, Notes for AIM-Workshop on spectra of families of
matrices described by graphs, digraphs and sign patterns, 2006.
[3] I. Gutman, The Energy of a Graph, Ber. Math. Stat. Sekt. Forschungszentrum Graz,
103, (1978) 1-22.
[4] I. Gutman, The energy of a graph : old and new results, in Algebraic Combinatorics and
Applications, A.Betten, A. Kohner, R. Laue and A.Wassermann, eds., Springer, Berlin,
(2001) 196-211.
[5] I. Gutman and Polanski, Mathematical Concepts in Organic Chemistry, Springer-Verlag,
Berlin, 1986.
Math.Combin.Book Ser. Vol.1 (2010), 115-121

Euler-Savary’s Formula for the Planar Curves in

Two Dimensional Lightlike Cone

Handan BALGETİR ÖZTEKİN and Mahmut ERGÜT

(Department of Mathematics, Fırat University, 23119 Elazığ, TÜRKİYE)

Email: [email protected] & [email protected]

Abstract: In this paper, we study the Euler-Savary’s formula for the planar curves in the
lightlike cone. We first define the associated curve of a curve in the two dimensional lightlike
cone Q2 .Then we give the relation between the curvatures of a base curve, a rolling curve
and a roulette which lie on two dimensional lightlike cone Q2 .

Keywords: Lightlike cone, Euler Savary’s formula, Smarandache geometry, Smaran-

dachely denied-free.

AMS(2010): 53A04, 53A30, 53B30

§1. Introduction

The Euler-Savary’s Theorem is well known theorem which is used in serious fields of study in
engineering and mathematics.
A Smarandache geometry is a geometry which has at least one Smarandachely denied
axiom(1969), i.e., an axiom behaves in at least two different ways within the same space, i.e.,
validated and invalided, or only invalided but in multiple distinct ways. So the Euclidean
geometry is just a Smarandachely denied-free geometry.
In the Euclidean plane E 2 ,let cB and cR be two curves and P be a point relative to cR .
When cR roles without splitting along cB , the locus of the point P makes a curve cL. The curves
cB , cR and cL are called the base curve, rolling curve and roulette, respectively. For instance,
if cB is a straight line, cR is a quadratic curve and P is a focus of cR , then cL is the Delaunay
curve that are used to study surfaces of revolution with the constant mean curvature, (see
[1]).The relation between the curvatures of this curves is called as the Euler-Savary’s formula.
Many studies on Euler-Savary’s formula have been done by many mathematicians. For
example, in [4], the author gave Euler-Savary’s formula in Minkowski plane. In [5], they ex-
pressed the Euler-Savary’s formula for the trajectory curves of the 1-parameter Lorentzian
spherical motions.
On the other hand, there exists spacelike curves, timelike curves and lightlike(null) curves in
semi-Riemannian manifolds. Geometry of null curves and its applications to general reletivity
in semi-Riemannian manifolds has been constructed, (see [2]). The set of all lightlike(null)
1 Received February 1, 2010. Accepted March 30,2010.
116 Handan BALGETİR ÖZTEKİN and Mahmut ERGÜT

vectors in semi-Riemannian manifold is called the lightlike cone. We know that it is important
to study submanifolds of the lightlike cone because of the relations between the conformal
transformation group and the Lorentzian group of the n-dimensional Minkowski space E1n and
the submanifolds of the n-dimensional Riemannian sphere S n and the submanifolds of the
(n+1)-dimensional lightlike cone Qn+1 . For the studies on lightlike cone, we refer [3].
In this paper, we have done a study on Euler-Savary’s formula for the planar curves in
two dimensional lightlike cone Q2 . However, to the best of author’s knowledge, Euler-Savary’s
formula has not been presented in two dimensional lightlike cone Q2 . Thus, the study is proposed
to serve such a need. Thus, we get a short contribution about Smarandache geometries.
This paper is organized as follows. In Section2, the curves in the lightlike cone are reviewed.
In Section3, we define the associated curve that is the key concept to study the roulette, since
the roulette is one of associated curves of the base curve. Finally, we give the Euler-Savary’s
formula in two dimensional cone Q2 .
We hope that, these study will contribute to the study of space kinematics, mathematical
physics and physical applications.

§2. Euler-Savary’s Formula in the Lightlike Cone Q2

Let E13 be the 3−dimensional Lorentzian space with the metric

g(x, y) = hx, yi = x1 y1 + x2 y2 − x3 y3 ,

where x = (x1 , x2 , x3 ), y = (y1 , y2 , y3 ) ∈ E13 .

The lightlike cone Q2 is defined by

Q2 = {x ∈ E13 : g(x, x) = 0}.

Let x : I → Q2 ⊂ E13 be a curve, we have the following Frenet formulas (see [3])

x′ (s) = α(s)
′
α (s) = κ(s)x(s) − y(s) (2.1)
y ′ (s) = −κ(s)α(s),

where s is an arclength parameter of the curve x(s). κ(s) is cone curvature function of the
curve x(s),and x(s), y(s), α(s) satisfy

hx, xi = hy, yi = hx, αi = hy, αi = 0,

hx, yi = hα, αi = 1.

For an arbitrary parameter t of the curve x(t), the cone curvature function κ is given by
D E2 D 2 E
dx d2 x d x d2 x dx dx
,
dt dt 2 − dt2 , dt2 dt , dt
κ(t) = 5 (2.2)
2 dx dx
dt , dt

Using an orthonormal frame on the curve x(s) and denoting by κ, τ , β and γ the curvature,
the torsion, the principal normal and the binormal of the curve x(s) in E13 , respectively, we
 Euler-Savary’s Formula for the Planar Curves in Two Dimensional Lightlike Cone 117

have

x′ = α
′
α = κx − y = κβ,

where κ 6= 0, hβ, βi = ε = ±1, hα, βi = 0, hα, αi = 1, εκ < 0. Then we get

κx − y κ′ 1
β = ε√ , ετ γ = √ (x + y). (2.3)
−2εκ 2 −2εκ κ

Choosing r
−εκ 1
γ= (x + y), (2.4)
2 κ
we obtain
√ 1 κ′
κ= −2εκ, τ = − ( ). (2.5)
2 κ

Theorem 2.1 The curve x : I → Q2 is a planar curve if and only if the cone curvature function
κ of the curve x(s) is constant [3].

If the curve x : I → Q2 ⊂ E13 is a planar curve, then we have following Frenet formulas

x′ = α,
√
α′ = ε −2εκβ, (2.6)
′
√
β = − −2εκα.

Definition 2.2 Let x : I → Q2 ⊂ E13 be a curve with constant cone curvature κ (which means
that x is a conic section) and arclength parameter s. Then the curve

xA = x(s) + u1 (s)α + u2 (s)β (2.7)

is called the associated curve of x(s) in the Q2 , where {α, β} is the Frenet frame of the curve
x(s) and {u1 (s), u2 (s)} is a relative coordinate of xA (s) with respect to {x(s), α, β}.

Now we put
dxA δu1 δu2
= α+ β. (2.8)
ds ds ds
Using the equation (2.2) and (2.6), we get

dxA du1 √ √ du2

= (1 + − −2εκu2 )α + (u1 ε −2εκ + )β. (2.9)
ds ds ds
Considering the (2.8) and (2.9), we have

δu1 du1 √
= (1 + − −2εκu2 )
ds ds
δu2 √ du2
= (u1 ε −2εκ + ) (2.10)
ds ds
118 Handan BALGETİR ÖZTEKİN and Mahmut ERGÜT

Let sA be the arclength parameter of xA . Then we write

dxA dxA dsA
= . = v1 α + v2 β (2.11)
ds dsA ds
and using (2.8) and (2.10), we get

du1 √
v1 = 1+ − −2εκu2
ds
√ du2
v2 = u1 ε −2εκ + . (2.12)
ds
The Frenet formulas of the curve xA can be written as follows:
dαA √
= εA −2εA κA βA
dsA
dβA √
= − −2εA κA αA , (2.13)
dsA
where κA is the cone curvature function of xA and εA = hβA , βA i = ±1 and hαA , αA i = 1.
Let θ and ω be the slope angles of x and xA respectively. Then
dω dφ 1
κA = = (κ + )p 2 , (2.14)
dsA ds |v1 + εv22 |

where φ = ω − θ.
If β is spacelike vector, then we can write
v1 v2
cos φ = p 2 2
and sin φ = p 2 .
v1 + v2 v1 + v22

Thus, we get
dφ d v1
= (cos−1 p 2 )
ds ds v1 + v22
and (2.14) reduces to
′ ′
v1 v2 − v1 v2 1
κA = (κ + )p 2 .
v12 + v22 v1 + v22
If β is timelike vector, then we can write
v1 v2
cosh φ = p 2 2
and sinh φ = p 2
v1 − v2 v1 − v22

and we get
dφ d v1
= (cosh−1 p 2 ).
ds ds v1 − v22
Thus, we have
′ ′
v1 v2 − v1 v2 1
κA = (κ + )p 2 .
v12 − v22 v1 − v22
Let xB and xR be the base curve and rolling curve with constant cone curvature κB and
κR in Q2 , respectively. Let P be a point relative to xR and xL be the roulette of the locus of
P.
 Euler-Savary’s Formula for the Planar Curves in Two Dimensional Lightlike Cone 119

We can consider that xL is an associated curve of xB such that xL is a planar curve in Q2 ,

then the relative coordinate {w1, w2 } of xL with respect to xB satisfies
δw1 dw1 √
= 1+ − −2εB κB w2
dsB dsB
δw2 √ dw2
= w1 εB −2εB κB + (2.15)
dsB dsB
by virtue of (2.10).
Since xR roles without splitting along xB at each point of contact, we can consider that
{w1, w2 } is a relative coordinate of xL with respect to xR for a suitable parameter sR . In this
case, the associated curve is reduced to a point P. Hence it follows that
δw1 dw1 √
= 1+ − −2εR κR w2 = 0
dsR dsR
δw2 √ dw2
= w1 εR −2εR κR + = 0. (2.16)
dsR dsR
Substituting these equations into (2.15), we get
δw1 √ √
= ( −2εR κR − −2εB κB )w2
dsB
δw2 √ √
= (εB −2εB κB − εR −2εR κR )w1 . (2.17)
dsB
If we choose εB = εR = −1, then
δw1 2 δw2 2 √ √
0<( ) −( ) = ( 2κR − 2κB )2 (w22 − w12 ). (2.18)
dsB dsB
Hence, we can put
w1 = r sinh φ , w2 = r cosh φ.
Differentiating this equations, we get
dw1 dr dφ
= sinh φ + r cosh φ
dsR dsR dsR
dw2 dr dφ
= cosh φ + r sinh φ (2.19)
dsR dsR dsR
Providing that we use (2.16), then we have
dw1 √
= r 2κR cosh φ − 1
dsR
dw2 √
= r sinh φ 2κR (2.20)
dsR
If we consider (2.19) and (2.20), then we get
dφ √
r = −r 2κR + cosh φ (2.21)
dsR
Therefore, substituting this equation into (2.14), we have
cosh φ
rκL = ±1 + √ √ (2.22)
r 2κR − 2κB
120 Handan BALGETİR ÖZTEKİN and Mahmut ERGÜT

If we choose εB = εR = +1, then from (2.17)

δw1 2 δw2 2 √ √
0<( ) +( ) = ( −2κR − −2κB )2 (w12 + w22 ) (2.23)
dsB dsB
Hence we can put

w1 = r sin φ , w2 = r cos φ.
Differentiating this equations, we get
dw1 dr dφ √
= sin φ + r cos φ = r −2κR cos φ − 1
dsR dsR dsR
dw2 dr dφ √
= cos φ − r sin φ = −r sin φ −2κR (2.24)
dsR dsR dsR
and
dφ √
r = r −2κR − cos φ (2.25)
dsR
Therefore, substituting this equation into (2.14), we have
√ √
−2κB + −2κR cos φ
rκL = √ √ − √ √ , (2.26)
−2κR − −2κB r −2κR − −2κB
√
where κL = −2εL κL .
Thus we have the following Euler-Savary’s Theorem for the planar curves in two dimen-
sional lightlike cone Q2 .

Theorem 2.3 Let xR be a planar curve on the lightlike cone Q2 such that it rolles without
splitting along a curve xB . Let xL be a locus of a point P that is relative to xR . Let Q be a
point on xL and R a point of contact of xB and xR corresponds to Q relative to the rolling
relation. By (r, φ) , we denote a polar coordinate of Q with respect to the origin R and the base
′
line xB |R. Then curvatures κB , κR and κL of xB , xR and xL respectively, satisfies
cosh φ
rκL = ±1 + √ √ , if εB = εR = −1,
r 2κR − 2κB
√ √
−2κB + −2κR cos φ
rκL = √ √ − √ √ if εB = εR = +1.
−2κR − −2κB r −2κR − −2κB

References

[1] Hano, J. and Nomizu, K., Surfaces of Revolution with Constant Mean Curvature in
Lorentz-Minkowski Space, Tohoku Math. J., 36 (1984), 427-437.
[2] Duggal, K.L. and Bejancu, A., Lightlike Submanifolds of Semi-Riemannian Manifolds and
Applications, Kluwer Academic Publishers,1996.
[3] Liu, H., Curves in the Lightlike Cone, Contributions to Algebra and Geometry, 45(2004),
No.1, 291-303.
[4] Ikawa, T., Euler-Savary’s Formula on Minkowski Geometry, Balkan Journal of Geometry
and Its Applications, 8(2003), No. 2, 31-36.
 Euler-Savary’s Formula for the Planar Curves in Two Dimensional Lightlike Cone 121

[5] Tosun, M., Güngör, M.A., Okur, I., On the 1-Parameter Lorentzian Spherical Motions and
Euler-Savary formula, American Society of Mechanical Engineering, Journal of Applied
Mechanic, vol.75, No.4, 972-977, September 2007.
The man with a new idea is a crank until the idea succeeds.

By Mark Twain, an American writer.

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 the Mathematical Combinatorics (International
Book Series (ISBN 978-1-59973-120-9). An effort is made to publish a paper duly recom-
mended by a referee within a period of 3 months. Articles received are immediately put the
referees/members of the Editorial Board for their opinion who generally pass on the same in
six week’s time or less. In case of clear recommendation for publication, the paper is accom-
modated 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, Combinatorial speculation and combinatorial conjecture for mathematics, In-
ternational J.Math. Combin., Vol.1, 1-19(2007).
[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.
Mathematical Combinatorics (International Book Series) April, 2010

Contents

Singed Total Domatic Number of a Graph

BY H.B. WALIKAR, SHAILAJA S.SHIRKOL, KISHORI P.NARAYANKAR . . . . . . . . 01
Radio Number of Cube of a Path
BY B.SOORYANARAYANA, VISHU KUMAR M. and MANJULA K. . . . . . . . . . . . . . . 04
Biprimitive Semisymmetric Graphs on P SL(2, p)
BY FURONG WANG AND LIN ZHANG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .30
A Note on Path Signed Digraphs
BY P.SIVA KOTA REDDY, S.VIJAY AND H.C.SAVITHRI . . . . . . . . . . . . . . . . . . . . . . . . . 42
A Combinatorial Decomposition of Euclidean Spaces Rn with Contribution
to Visibility BY LINFAN MAO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
Counting Rooted Eulerian Planar Maps
BY JUNLIANG CAI AND YANPEI LIU . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
The nth Power Signed Graphs-II
BY P.SIVA KOTA REDDY, S.VIJAY AND V.LOKESHA . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
Dynamical Knot and Their Fundamental Group
BY M.ABU-SALEEM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
The Crossing Number of the Cartesian Product of Star Sn with a
6-Vertex Graph BY JING WANG AND YUANQIU HUANG . . . . . . . . . . . . . . . . . . . . 87
Smarandaches Pedal Polygon Theorem in the Poincaré Disc Model
of Hyperbolic Geometry BY CĂTĂLIN BARBU . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
The Arc Energy of Digraph
BY Z.KHOSHBAKHT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
Euler-Savarys Formula for the Planar Curves in Two Dimensional
Lightlike Cone BY HANDA BALGETİR ÖZTEKİN AND MAHUMT ERGÜT . . . 115

ISBN 9781599731209

9 781599 731209