Int. J. Contemp. Math. Sciences, Vol. 4, 2009, no.

10, 467 - 474

An Introduction to Order Prime Graph

M. Sattanathan* and R. Kala**

Department of Mathematics, Manonmaniam Sundaranar University

Tirunelveli- 627 012, Tamil Nadu, India
* satta [email protected]
** [email protected]

Abstract

Let Γ be a finite group. The order prime graph OP (Γ) of a group Γ

is a graph with V (OP (Γ))=Γ and two vertices are adjacent in OP (Γ)
if and only if their orders are relatively prime in Γ. In this paper we
obtain several properties of OP (Γ), lower bounds and upper bounds on
the number of edges of OP (Γ) and characterise certain classes of order
prime graph.

Mathematics Subject Classification: 05C

Keywords: Order prime graph, unicyclic graph, triangulated graph,

generating set

1 INTRODUCTION
Thoughout this paper, Γ denotes a group with identity e and G denotes a
graph with order p and size q. Terms not defined here are used in the sense of
Apostol1 ,Harary 3 and Herstein4 .
Two integers a and b are said to be relatively prime if their greatest common
divisor is 1 viz., (a, b)=1. Relatively prime integers play a significant role in
both Analytic and Algebraic number theory. They motivated us to define order
prime graph OP (Γ), where Γ is a finite group. We hope that this definition
will be a foundation stone for a new development in Algebraic Graph Theory.
It is defined as a graph with V (OP (Γ))=Γ , where Γ is a finite group and two
vertices are adjacent in OP (Γ) if and only if their orders are relatively prime
in Γ.
468 M. Sattanathan and R. Kala

Definition 1.1. [3] Unicyclic graphs are graphs which are connected and have
just one cycle.

Definition 1.2. [5] A graph G in which each cycle of length at least four has
a chord is called a triangulated graph.

Definition 1.3. [2] If A is any subset of a group G, the subgroup of G gener-

ated by A, [A] is defined as the intersection of all subgroups of G containing
A.
If [A]=G, then A is said to be a generating set of G.

We use the following theorems.

Theorem 1.4. (CAUCHY)[4] If p is a prime number and p |o(G) , then G has

an element of order p.

Theorem 1.5. [4] If H and K are finite subgroups of G with orders o(H) and
o(K) respectively, then o(HK)= o(H)o(K)
o(H∩K)
.

2 MAIN RESULTS
Definition 2.1. Let Γ be a finite group. The order prime graph OP (Γ) of a
group Γ is a graph with V (OP (Γ)) = Γ and two vertices a and b are adjacent
in OP (Γ) if and only if (o(a), o(b)) = 1. Here o(a),o(b) respectively denote the
orders of a and b .

Example 2.2. Let Γ=Z6. Then (Γ, +6 ) is a group where +6 is addition mod-
ulo 6. The graph OP (Γ) is given in Figure 1.
We observe that o(0) = 1, o(1) = 6, o(2) = 3, o(3) = 2, o(4) = 3, o(5) = 6.
1s 0s 5s
@
@
@
@
@
s s @s
2 3 4
Figure 1: OP (Γ)

Proposition 2.3. Let Γ be a group with o(Γ)=n. Then Δ(OP (Γ))=n − 1.

Introduction to order prime graph 469

Proof. Identity is the only element of order 1 in Γ and hence the proof follows.

Proposition 2.4. For any group Γ, OP (Γ) is complete if and only if o(Γ)=2.

Proof. If o(Γ)=2, clearly OP (Γ)∼ =K2 . Suppose OP (Γ) is complete. If o(Γ)

≥ 3 and every element in Γ is a self-inverse element, then every element is of
order 2 and so OP (Γ) cannot be complete. If there exists at least one element
which is not a self-inverse element , the element and its inverse have the same
order in Γ and hence non-adjacent in OP (Γ). These contradictions show that
o(Γ)=2.

Corollary 2.5. Let Γ be a group with n≥ 3 elements. Then OP (Γ) cannot be

regular.

Proof. The proof follows from proposition 2.3 and proposition 2.4.

Proposition 2.6. For any group Γ, G=OP (Γ) can never be a unicyclic graph.

Proof. Suppose G is unicyclic. Since Δ(G)=n − 1 and e is the vertex with

degree Δ,G − e has exactly one edge e . Let e =ab. Then (o(a), o(b))=1. Now
at least one of a and b should not be a self-inverse element since otherwise
o(a)=o(b)=2. Let a= a−1 . Now o(a)=o(a−1 ) and so a−1 , b are also adjacent
in G − e, which is a contradiction.

Theorem 2.7. Let Γ be a group with n elements. OP (Γ) ∼

= K1,n−1 if and
+
only if o(Γ)=p where p is a prime number and α∈ Z .
α

Proof. Let Γ={e, a1 , ..., an−1 }. Assume o(Γ)= pα where p is a prime number
and α ∈ Z+ . Then ∀i o(ai )=pk for some k ∈ {1, 2, ..., α}. Now ai s are mutually
non adjacent in OP (Γ) and hence OP (Γ)∼ = K1,n−1 .
∼
Conversely, assume that G=OP (Γ) = K1,n−1 . Clearly G − e is totally discon-
nected.
Claim: o(Γ)=pα .
Suppose o(Γ) = pα . Without loss of generality we shall assume that
o(Γ)=p1 k1 p2 k2 where p1 , p2 are prime numbers and k1 ,k2 ∈ Z+ . Since p1 |o(Γ)
and p2 |o(Γ), by Theorem 1.4, Γ has elements of order p1 and p2 . Let a, b be
elements of Γ  o(a)=p1 and o(b)=p2 . Hence a, b are adjacent in G − e, which
is a contradiction.

Corollary 2.8. Let Γ be a group with n elements. OP (Γ) is a tree if and only
if o(Γ)=pα where p is a prime number and α ∈ Z+ .
470 M. Sattanathan and R. Kala

Proof. Since Δ(OP (Γ))=n − 1, OP (Γ) is connected. Hence OP (Γ) is a tree

⇔ OP (Γ) ∼
= K1,n−1 ⇔ o(Γ)=pα where p is prime and α ∈ Z+ .

Proposition 2.9. Let Γ be a cyclic group. Then OP (Γ) has at least two
pendent vertices.

Proof. If Γ is a cyclic group of order two , then OP (Γ)∼ = K2 and hence the
proof follows.Suppose Γ is a cyclic group of order greater than two. Let a
be the generator of Γ. Clearly o(a)=o(Γ). Since o(Γ) = 2, a−1 = a is also a
generator of Γ and so o(a)=o(a−1 )=o(Γ). Since the order of any element of Γ
divides the order of Γ, a and a−1 are adjacent to identity element only. Thus
OP (Γ) has at least two pendent vertices.

Remark 2.10. OP (Γ) need not always be a triangulated graph. Let Γ1 ∼ =

S3 . OP (Γ1) is given in Figure 2. The graph induced by the set
{(12), (23), (123), (132)} in OP (Γ1) is C4 and so OP (Γ1) is not a triangulated
graph. But the graph given in example 2.2 is a triangulated graph.
(12)
s es (13)
s
H 
HH @ 
HH  @
H @
H
  HH@
 H@
s
 s H@s
H
(123) (23) (132)
Figure 2: OP (Γ1)

Theorem 2.11. If Γ1 , Γ2 are two groups such that Γ1 ∼

=Γ2 , then
∼
OP (Γ1)=OP (Γ2). But the converse is not true.

Proof. Let Φ be an isomorphism of Γ1 onto Γ2 .

Now, a and b are adjacent in OP (Γ1)
⇔(o(a), o(b)) = 1
⇔(o(Φ(a)), o(Φ(b)))=1
⇔ Φ(a) and Φ(b) are adjacent in OP (Γ2). Thus OP (Γ1)∼ =OP (Γ2).
Conversely, consider the groups (Γ1 , .), (Γ2 , Δ) where Γ1 ={1, −1, i, −i},
Γ2 ={Φ, {a} , {b} , {ab}} and ·, Δ respectively denote multiplication and sym-
metric difference. Since o(OP (Γ1))=o(OP (Γ2))=22 , by Theorem 2.7, OP (Γ1)
∼
= OP (Γ2) ∼= K1,3 . But Γ1 has an element of order 4 whereas Γ2 does not have
an element of order 4. Thus Γ1 ∼ =Γ2 .

Theorem 2.12. Let Γ be a group. Then Aut(Γ) ⊆ Aut(OP (Γ))

Introduction to order prime graph 471

Proof. Let Γ be a group . g ∈ Aut(Γ)

⇒ g is an isomorphism of Γ onto Γ. Now, a and b are adjacent in OP (Γ)
⇔ (o(a), o(b)) = 1
⇔ (o(g(a)), o(g(b))) =1
⇔ g(a) and g(b) are adjacent in OP (Γ). Hence Aut(Γ) ⊆ Aut(OP (Γ))
Remark 2.13. The converse of theorm 2.12 is not true. Consider the group
(Γ = Z5 , +5 ). Here OP (Γ) ∼
= K1,4 .Define g:OP (Γ) → OP (Γ) such that g(0)=0,
g(1)=2, g(2)=3, g(3)=4, g(4)=1. Clearly g is an automorphism of OP (Γ).But
g(1 +5 2)=g(3)=4 and g(1) +5 g(2) =2 +5 3=0.Hence g(1 +5 2) = g(1) +5 g(2)
so that g is not an automorphism of Γ.
Theorem 2.14. Let Γ be a group. Suppose OP (Γ) has two adjacent elements
a, b such that o(a)o(b)=o(Γ). Then the set {a, b} is a generating set of Γ.
Proof. Given a, b are two adjacent elements in OP (Γ) and so (o(a), o(b))=1.
Let o(a)=m,o(b)=n. Let H=< a > and K=< b > be two subgroups of Γ. We
have o(H)=m and o(K)=n. Since (m, n)=1 , H ∩ K={e},
o(H ∩ K)=1. By Theorem 1.5 o(HK)= o(H)o(K)
o(H∩K)
=o(H)o(K)=mn=o(Γ).
Hence o(HK)=o(Γ). Since HK is a subset of Γ,HK=Γ. a, b ∈ Γ ⇒ [{a, b}]
⊆ Γ. Now x ∈ HK ⇒ x=an bm ⇒ x ∈ [{a, b}].
Γ = HK ⊆ [{a, b}]. Γ = [{a, b}] and hence {a, b} is a generating set of Γ.
Theorem 2.15. Let Γ be an abelian group. Let X ⊆ Γ be such that the graph
induced by X is complete in OP (Γ) and the product of the order of all elements
of X is same as o(Γ). Then X is a generating set of Γ.
Proof. Let X={a1 , a2 , ..., ak } ⊆ Γ. By assumption, < X > is complete in
OP (Γ) and so (o(ai ), o(aj ))=1 ∀ i = j. For 1≤ i ≤ k, consider the subgroups
of Γ, Hi =< ai >. Since (o(Hi ), o(Hj ))=1 ∀ i = j, Hi ∩ Hj ={e} ∀ i = j. Now
1 )o(H2 ...Hk )
o(H1 H2 ...Hk )= o(H
o(H1 ∩H2 ...Hk )
. Since (H1 ∩ H2 )(H1 ∩ H3 )...(H1 ∩ Hk ) ⊆ H1 ∩
o(H1 )o(H2 ...Hk )
H2 ...Hk , o(H1 H2 ...Hk ) ≥ o((H1 ∩H2 )(H1 ∩H3 )...(H1 ∩Hk ))
= o(H1 )o(H
1
2 ...Hk )
=o(H1 )o(H2 ...Hk )
1 )o(H2 )o(H3 ...Hk )
= o(Ho(H 2 ∩H3 ...Hk )
.
Proceeding like this we get, o(H1 H2 ...Hk ) ≥ o(H1 )o(H2 )...o(Hk )=o(Γ). Hence
H1 H2 ...Hk =Γ. Also a ∈ H1 H2 ...Hk ⇒ a=ar11 ar22 ...arkk ⇒
a ∈ [{a1 , a2 , ..., ak }]. Hence H1 H2 ...Hk ⊆ [{a1 , a2 , ..., ak }].
So Γ =[{a1 , a2 , ..., ak }]. Hence the set X={a1 , a2 , ..., ak } is a generating set of
Γ.
Theorem 2.16. Let Γ be a group with o(Γ)= pn1 1 pn2 2 ...pnk k where pi s are prime
numbers and ni ∈ Z+ (1≤ i ≤ k). Then OP (Γ) is a complete (k + 1)-partite
graph if and only if o(a) =pij ∀ a ∈ Γ − e, 1 ≤ i ≤ k and 1 ≤ j ≤ ni .
472 M. Sattanathan and R. Kala

Proof. Let Γ be a group with o(Γ)=pn1 1 pn2 2 ...pnk k where pi s are prime numbers
and ni ∈ Z+ (1≤ i ≤ k). For every 1≤ i ≤ k, define the set
Vi ={a ∈ Γ − e/o(a) = pi j , 1 ≤ j ≤ ni } and V0 = {e}. Theorem 1.4 gaurantees

that Vi = Φ. Also Vi ∩ Vj =Φ ∀ i = j and ki=0 Vi =Γ. No two elements in
Vi are adjacent and (pi r , pj s )=1 ∀ i = j , 1 ≤ r ≤ ni and 1≤ s ≤ nj . Hence
OP (Γ) is a complete (k + 1)-partite graph with parition V0 , V1 , ..., Vk .
Conversely, assume that OP (Γ) is a complete (k + 1)-partite graph. Clearly
one partition is the single element e. Without loss of generality we shall assume
that it is the first partition.According to theorem 1.4 Γ has elements ai such
that o(ai )=pi , 1≤ i ≤ k. No two ai s can lie in the same partition of OP (Γ).
Let us suppose that ai belongs to (i + 1)th partition. Each (i + 1)th -partition
has the elements of order mpi where m ∈ Z+ .
Claim: o(a)=pi j ∀ a ∈ Γ − e, 1≤ j ≤ ni and 1≤ i ≤ k.
If not , ∃ an element b ∈ Γ − e  o(b) = pi j . Without loss of generality we
shall assume that o(b)=p1 p2 . Hence b lies in either 2nd or 3rd partition and
accordingly a2 b or a1 b is not an edge. This is a contradiction as OP (Γ) is a
complete (k + 1)-partite graph.

Theorem 2.17. Let Γ be a group with o(Γ)=n=p1 n1 p2 n2 ...pk nk . Let q be the

number of edges of the graph G=OP (Γ).Then n − 1 + 12 [m2 − (m1 2 + ... + mk 2 )]
≤ q ≤ 12 [n2 − n + m − (m1 2 + ... + mk 2 )].....(1) where mi = the number of
elements of Γ of order pi j 1≤ j ≤ ni ,1 ≤ i ≤ k, and m=m1 + m2 + ... + mk .
Moreover, these bounds are sharp.

Proof. Clearly the identity element is adjacent to remaining (n − 1) elements,

and each of the mi elements are adjacent to all the mj elements whenever i =
j. Hence the minimum number of edges in G equals
n − 1 + 12 [m1 (m − m1 ) + m2 (m − m2 ) + ... + mk (m − mk )]
=n − 1 + 12 [mm1 − m1 2 + ... + mmk − mk 2 ]
=n − 1 + 12 [m(m1 + m2 + ... + mk ) − (m1 2 + ... + mk 2 )]
=n − 1 + 12 [m2 − (m1 2 + ... + mk 2 )]. Also these mi elements are mutually
non-adjacent. Hence the maximum number of edges of G equals
n m1  m2  mk 
2
- 2 +  2
+ ... + 2 
m1 (m1 −1) mk (mk −1)
= n(n−1)
2
− 2
+ ... +
2
1 2
= 2 [n − n − {m1 2 + 2
... + mk − (m1 + ... + mk )}]
1 2
= 2 [n − n + m − (m1 2 + ... + mk 2 )]. Hence n − 1 + 12 [m2 − (m1 2 + ... + mk 2 )]
≤ q ≤ 12 [n2 − n + m − (m1 2 + ... + mk 2 )].
Moreover, the following examples exhibit that the bounds are sharp. Consider
Γ given in example 2.2. Here n=6,m1 =2,m2 =1 and m=3. The number of
Introduction to order prime graph 473

edges in OP (Γ)=(6-1)+ 12 [32 − (22 + 12 )]=5+2=7. If one considers OP (Γ1)

given in the Figure 2, then n=6, m1 =3, m2 =2 and m=5. The number of edges
in OP (Γ1)= 12 [62 − 6 + 5 − (32 + 22 )]= 12 [22]=11.

Corollary 2.18. Let Γ be a group with o(Γ)=n=p1 n1 p2 n2 ...pk nk . Let q be the

number of edges of the qraph OP (Γ). Then the upper bound and lower bound
of q in (1) are equal if and only if OP (Γ) is a complete (k + 1)-partite graph.

Proof. Assume that the upper bound and lower bound of q in (1) are equal.
Thus we have n−1+ 12 [m2 −(m1 2 +...+mk 2 )] = 12 [n2 −n+m−(m1 2 +...+mk 2 )]
⇒ 2n − 2 + m2 =n2 − n + m
⇒ m2 − m=n2 − n − 2n + 2
⇒ m2 − m=(n − 2)(n − 1)
⇒ m2 − m − [(n − 1)(n − 2)]=0
⇒ (m + (n − 2))(m + (1 − n))=0
⇒ m=−(n − 2) or m=n − 1. But m=−(n − 2) is not possible. Hence m=n − 1.
This means o(a)=pi j ∀ a ∈ Γ − e, 1 ≤ j ≤ ni and 1≤ i ≤ k. Thus by
Theorem 2.16 OP (Γ) is a complete (k + 1)-partite graph. Conversely assume
that OP (Γ) is a complete (k + 1)-partite graph. By Theorem 2.16 o(a)=pi j ∀
a ∈ Γ − e, 1 ≤ j ≤ ni and 1≤ i ≤ k. Hence m=n − 1. Now lower bound of q
=n − 1 + 12 [m2 − (m1 2 + ... + mk 2 )]
=n − 1 + 12 [(n − 1)2 − (m1 2 + ... + mk 2 )]
=n − 1 + 12 [n2 − 2n + 1 − (m1 2 + ... + mk 2 )]
= 12 [2n − 2 + n2 − 2n + 1 − (m1 2 + ... + mk 2 )]
= 12 [n2 − 1 − (m1 2 + ... + mk 2 )].
Upper bound of q= 12 [n2 − n + m − (m1 2 + ... + mk 2 )]
= 12 [n2 − n + n − 1 − (m1 2 + ... + mk 2 )]
= 12 [n2 − 1 − (m1 2 + ... + mk 2 )]. Hence the lower bound of q equals the upper
bound of q.

References
[1] Apostol,T.M.,Introduction to Analytic number theory,Narosa publishing
house,1998.

[2] Dipak Chatterjee,Abstract Algebra,Prentice Hall of india private lim-

ited,2001.

[3] F.Harary,Graph Theory,Addison Wesley,Reading,Mass,1972.

474 M. Sattanathan and R. Kala

[4] I.N.Herstein,Topics in Algebra, Second edition,John Wiley & Sons,2003.

[5] K.R.Parthasarathy,Basic Graph Theory,Tata Mcgraw-Hill publishing

Company Limited,1994.

Received: August, 2008