Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

Towards a Theory of Domination

in Graphs
E. J. Cockayne
University of Victoria
Victoria, B.C., Canada
S. T. Hedetniemi
University of Oregon
Eugene, Oregon

ABSTRACT

This paper presents a quick review of results and applica-


tions concerning dominating sets i n graphs. The domatic number
of a graph is defined and studied. I t is seen that the theory
of domination resembles the we22 knuwn theory of colorings of
graphs.
1. INTRODUCTION

A s e t of vertices D i n a graph G = ( V I E ) i s a dominating


s e t i f every v e r t e x n o t i n
D i s a d j a c e n t t o a t least one v e r t e x
i n D. A s e t of vertices D is independent i f no t w o v e r t i c e s i n
D are adjacent.
The concepts of domination and independence have e x i s t e d
f o r a long t i m e , two e a r l i e r d i s c u s s i o n s may be found i n O r e
[19, Ch.131 and Berge 121. A more up-to-date d i s c u s c i o n , and
perhaps the most comprehensive, on t h e subjects of +ndependence
and domination can be found i n Berge 131. W e n o t e t h a t Berge
h a s used the terms stable o r interiorly stable f o r what w e c a l l
independent and h a s used absorbant o r externaZZy stable f o r our
term dominating.
The l i t e r a t u r e c o n t a i n s many papers d e a l i n g w i t h t h e theory
of independent sets and t h e related t o p i c of graph colouring.
However, t h e r e are few papers about domination i n graphs, which
f a c t is somewhat s u r p r i s i n g p a r t i c u l a r l y i n view of t h e diver-
s i t y of a p p l i c a t i o n s . A b r i e f survey of t h e l i t e r a t u r e r e v e a l s
t h e following sample of a p p l i c a t i o n s f o r t h e concept of a domi-
n a t i n g set.
I n [19, p.2071, Ore mentions t h e problem of p l a c i n g a m i n i -
mum number of queens on a chessboard so t h a t each square is
c o n t r o l l e d by a t l e a s t one queen.

Networks, 7: 247-261
@ 1977 by John Wiley & Sons, Inc. 247
248 COCKAYNF, AND HEDETNIEMI

Berge [3, p.3031 mentions the problem of keeping a l l p o i n t s


i n a network under surveillance by a s e t of radar s t a t i o n s . In
a s i m i l a r vein, Liu [15, p.2351 discusses the application of dom-
inance t o communications i n a network, where a dominating s e t
represents a s e t of c i t i e s which, a c t i n g a s transmitting s t a t i o n s ,
can transmit messages t o every c i t y i n the network.
Berge a l s o discusses 13, p.3081 the r e l a t i o n s h i p between
kernels i n graphs, i . e . dominating s e t s which a r e a l s o independent
s e t s , and ' s o l u t i o n s ' i n game theory.
The notion of dominance i s a standard one i n coding theory.
I f one defines a graph whose v e r t i c e s a r e the n-dimensional vec-
t o r s with coordinates chosen from ( l , . . . , ~ and
) two v e r t i c e s a r e
adjacent i f they d i f f e r i n one coordinate, then sets of vectors
which a r e (n,p)-covering s e t s , s i n g l e e r r o r correcting codes,
o r p e r f e c t covering s e t s a r e a l l dominating s e t s of t h e graph
with c e r t a i n additional properties. See f o r example Kalbfleisch,
Stanton and Horton 1141.
The notion of dominance i s a l s o r e l a t e d t o the much studied
theory of matchings. A matching i s usually defined a s a maximal
s e t of independent edges i n a graph. Clearly, any matching i n a
graph G corresponds t o an independent dominating s e t i n the l i n e
graph L(G) of G.
Dominating s e t s a l s o occur i n applications of t h e recent
work of Edmonds and Fulkerson [lo] and B i l l e r a [41 on c l u t t e r s
and blockers of c l u t t e r s . A c l u t t e r i s a non-empty family C of
subsets of a s e t A, none of which properly contains any other.
The blocker o f a c l u t t e r c c o n s i s t s of a l l minimal subsets of A
which contain a t l e a s t one element of each subset i n C.
A clique of a graph G i s a maximal complete subgraph of G.
Clearly, the c o l l e c t i o n of cliques of a graph G forms a c l u t t e r ,
the blocker of which contains only dominating s e t s .
F i n a l l y , it i s c l e a r t h a t the s e t of a l l edges of a connected
graph G forms a c l u t t e r of the vertex s e t , the blocker of which
a l s o c o n s i s t s of dominating sets; i n p a r t i c u l a r the minimal sets
of v e r t i c e s which cover a l l edges of F ( a l s o c a l l e d minimal point
covers, c f . Harary 112, p.941).
In t h i s paper w e w i l l begin t o develop a theory of domina-
t i o n i n graphs. I n p a r t i c u l a r , we w i l l introduce t h e concept of
the domatic number of a graph and we w i l l obtain f o r t h i s para-
meter a number of r e s u l t s , some of which a r e s i m i l a r t o those
which have been developed f o r the chromatic number.

2. DEFINITIONS AND " 4 I N O L O G Y

The paper w i l l contain the following notations and terminol-


ogy. Any undefined term i n the following s e c t i o n s may be found
i n Harary 1121.
DOMINATION I N GRAPHS 249

For X,Y s u b s e t s of V ( G ) , X dominates Y in G i f f o r a l l y E Y


either y E X o r some v e r t e x o f X i s a d j a c e n t t o y i n G.
A D-partition of G i s a p a r t i t i o n o f V ( G ) i n t o dominating
sets.
The j o i n o f t w o graphs G , H i s t h e graph G + H = ( V ( G ) u V ( H ) , E )
where E = E ( G ) U E ( H I U { [ u , v ] lueV(G) , v s V ( H ) 1.
d (v) denotes t h e degree of t h e v e r t e x v i n t h e graph G.
G
x , t h e chromatic number of G, is t h e minimum o r d e r of a
(G)
p a r t i t i o n of V ( G ) i n t o independent s u b s e t s .
d ( G ) , t h e domatic nwnber of G, i s t h e maximum o r d e r of a D-
p a r t i t i o n o f G.
E ( G ) i s the maximum number of end edges i n a spanning f o r e s t
of G.
i d ( G ) , t h e idomatic number of G, i s t h e maximum o r d e r of a
p a r t i t i o n of V ( G ) i n t o sets which are b o t h independent and domi-
nating. ( W e n o t e t h a t such a p a r t i t i o n may n o t e x i s t i n which
case i d (G) = 0. )
e ( G ) i s t h e minimum o r d e r of a p a r t i t i o n of V ( G ) i n t o sub-
sets whose induced subgraphs i n G are complete.
i ( G ) ( B ( G I ) w i l l denote t h e smallest (largest) number of
0
v e r t i c e s i n a maximal independent s e t i n G.
c (GI ( C ( G ) w i l l denote t h e s m a l l e s t ( l a r g e s t ) number of
v e r t i c e s i n a c l i q u e o f G.
y ( G ) ( I ' ( G ) ) w i l l denote t h e smallest ( l a r g e s t ) number of
v e r t i c e s i n a minimal dominating s e t of G.
6 ( G ) ( A ( G ) ) w i l l denote t h e minimum ( m a x i m u m ) degree among
t h e vertices of G.
F i n a l l y , a graph G i s uniquely n-colorable i f x ( G ) = n and
t h e r e e x i s t s a unique p a r t i t i o n of V ( G ) i n t o n independent s e t s .

3. BRIEF SUMMARY OF E X I S T I N G RESULTS ON DOMINATING SETS

While making no c l a i m s t o be complete, w e survey i n t h i s


s e c t i o n known r e s u l t s i n v o l v i n g dominating sets. O t h e r refer-
ences were given i n t h e i n t r o d u c t i o n .
I n [19, 01.131 O r e s t a t e d five elementary p r o p e r t i e s of
minimal dominating sets, among which were t h e f o l l o w i n g two.

Proposition 3 . 2 : (Ore) A dominating s e t D is a minima2 domi-


nating s e t if and on22 if f o r each d E D one of the fo2lowing two
conditions holds:
1. d is not adjacent t o any vertex in D, or
2. there is a vertex c , d D such that N(c) n D = {dl.
250 COCKAYNE AND HEDETNIEMI

Proposition 3.2: (Ore) If G is a graph oithout isolated ver-


tices and D is a minimal dominating set, then V - D contains a
minimal dominating set.
I n 13, p.3091 Berge p o i n t s o u t t h e following property of
independent dominating s e t s .

Proposition 3.3: . (Berge) If D i s an independent dominating


set, then D i s both a minimal dominating s e t and a maximal in-
dependent set. ConverseZy, if D is a mdrnaz independent s e t
then D is an independent dominating set.
The following two elementary bounds f o r t h e domination
number y(G) a r e a l s o given i n [3, p. 3041.

Proposition 3 . 4 : (Berge) If G is a graph having p vertices,


q edges and m a x i m degree A, then

Vizing [203 a l s o has a r e s u l t r e l a t i n g p , q and y ( G ) .

Theorem 3.5: (Vizing) If G i s a graph having p vertices and


q edges, then \

q [Ip - yIG)) Ip - y(G) + 21/21.

In [ill, Gupta focused a t t e n t i o n on a v a r i e t y of r e l a t i o n -


ships among s e v e r a l parameters of graphs and mentioned t h e
obvious f a c t t h a t t h e minimum number of edges which a r e needed
t o cover a l l t h e edges of a graph G equals t h e domination num-
b e r of t h e l i n e graph of G.
In [71, i ( G ) , which may be c a l l e d t h e independent dominat-
ing number (see Proposition 3.31, w a s shown t o be r e l a t e d t o a
number of o t h e r e s t a b l i s h e d p a r t i t i o n parameters of graphs, in-
cluding a r b o r i c i t y , thickness and chromatic number. I n p a r t i c -
u l a r , it was shown t h a t i f one considers t h e i n t e r s e c t i o n graph
on the s e t of a l l independent s e t s i n a graph, denoted I ( G ) ,
then i ( I (G) = x ( G ) .
M. Aigner [l] has examined y(G) f o r s e v e r a l s p e c i a l c l a s s e s
of graphs which a r e s t u d i e d i n t h e theory of block designs.
Among o t h e r c l a s s e s , he has considered y ( G ) f o r t h e c l a s s of
graphs LR , t h e v e r t i c e s of which correspond t o t h e ordered r
rrn
t u p l e s of a s e t S with n elements, where two v e r t i c e s a r e adja-
c e n t i f they have a t l e a s t 2 coordinates i n common.
DOMINATION I N GRAPHS 251

Theorem 3.6: (Jaegm and P a y a n ) For a n y gmph G with p vertices

and

A by-product of t h e i r elegant proof of t h i s theorem i s the


following r e s u l t .

CoroZZaq 3.6: (Jaegar and Payan) For any graph G, y (z)-< d(G).
J. Nieminen 1171 has established the following equality f o r
the domination number of a graph.

Theorem 3:7: (Nierrrfnen) For a n y gmph G, with p vertices

Nieminen used the above r e s u l t t o e s t a b l i s h upper and lower


bounds f o r y ( G ) i n terms of a simply constructed spanning f o r e s t
of G.
F i n a l l y i n [81, the following question was answered: Given
integers p and d, what i s the m i n i m u m number of edges q i n a
graph having p v e r t i c e s and domatic number d?

4. PROPERTIES OF THE DOMATIC NUMBER

I t can be seen t h a t a D-partition i s a p a r t i t i o n of the ver-


t i c e s of G i n t o n subsets such t h a t every vertex is adjacent t o
a t l e a s t one vertex i n every subset other than i t s own. One way
of obtaining a D-partition of a graph i s t o assign colors t o the
v e r t i c e s of G i n such a way t h a t every vertex i s adjacent with a
vertex of every color d i f f e r e n t from i t s own. Thus, i f a graph
has domatic number k then it is evident t h a t every vertex must
be adjacent t o a t l e a s t k - 1 v e r t i c e s , one i n each dominating
subset of a D-partition of order k. Hence the following upper
bound f o r the domatic number i s obtained.

Proposition 4.1: For any gmph G, d(G) - 6(G) + 1.

W e w i l l say t h a t a graph G i s (domaticaZZy) fuzz i f


d(G) = 6(G) + 1.
I n the following proposition w e summarize a number of ele-
mentary r e s u l t s which determine d(G) f o r special classes of
graphs; the proofs of these r e s u l t s a r e simple and a r e omitted;
a s t e r i s k s indicate t h a t every graph i n the c l a s s i s f u l l .
252 COCKAYNE AND HEDETNIEMI

Proposition 4.2:
( i ) d(Kn + G) = n + d(G).

*(ii)d(Kn) = n; d(Fn) = I.

(iii) (Ore) d ( G ) 2 i f and only i f G has no


isolated veFtices.
* ( i v ) For any t r e e with p - 2 v e r t i c e s , d ( T ) = 2.
( v ) For any n - 1, *d(C3n) = 3, and

( v i ) For any 2 < n, d(K


m - 1 = m.
m, n
( v i i ) I f G i s uniquely n-colorable then d(G) 2 n.
* ( v i i i ) I f G i s ma7:mal outerplanar then d(G) = 3.
It is interesting to observe that in [ 6 ] a close relation-
ship was established between totally disconnected, acyclic, out-
erplanar and planar graphs. We have just shown that the (maximal)
totally disconnected and the maximal outerplanar graphs are all
full. However, the graph F6 (i.e. K with a 1-factor removed) is
6
an example of a maximal planar graph having 6 ( F 1 = 4 and d(F =3.
6 6
Thus, the maximal planar graphs are not all full.
We now establish a Nordhaus-Gaddum-type result for the
domatic number.

Proposition 4.3: For any graph G, having p v e r t i c e s ,

d(G) + d(?) -
< p + 1.

Proof: BY Proposition 4.1,

=p+l.

The next result sharpens Proposition 4.3.


DOMINATION I N GRAPHS 253

Theorem 4.4: Let G have p vertices, then d(G) + d(E) = p + 1


i f and only if G = K or
P P'
Proof: Certainly d(K 1 + d ( z = p + 1.
P P
W e proceed by induction on p. Suppose, contrary t o t6e
a s s e r t i o n , t h a t G # K o r K has p v e r t i c e s and d(G) + d (E) = p + 1.
P P
Case I: I f 6 ( G ) = 0, then d(G) = 1, G has an i s o l a t e d
v e r t e x and has a v e r t e x u of degree p - 1. Hence = K + F,
1
where F has p - 1 v e r t i c e s and F # Kp-1' Since d(F) < p - 1,

d(G) + d(G) = d(G) + 1 + d(F) (by Proposition 4.2 (i))


= 2 + d(F)
< p + l .
Case 2:
0 < 6 ( G ) < p/2
BY assumption,
d(a + d(G) = P + 1.
Therefore ,
d = p + 1 - d(G)
-
> (p + 1) - (6(G) + 1) (by Proposition 4.1)
= p - &(GI. (21
If a l l dominating s e t s i n a maximum D-partition of have a t
l e a s t two v e r t i c e s then
P -> 2 d(E) 1 2 ( p - 6(G)) (using ( 2 ) 1
'2P-P ( f r o m (1))
= P
a contradiction. Hence some vertex v dominates E.
Theref o r e ,
d-(v) = p
G
- 1 and dG(v) = 0.

Hence ,
6 ( G ) = 0. (contrary t o (1))
254 COCKAYNE AND HEDETNIEMI

Case 3:
p/2 5 < P - 1

In this case, since for any graph of p vertices

6(G) + 6(E)L p - 1,
6(z) L p - 1 - 6 ( G )
LP - 1 - P/2 (using ( 3 ) 1
= p/2 - 1.

If 6 (c) = 0, apply Case 1 to G, otherwise apply Case 2 to E.


This completes t h e proof.

As stated in Corollary 3 . 6 , the domination number of E is a


lower bound for the domatic number of G. We note, using Propo-
sitions 3 . 4 and 4.1, that these two parameters have the same
upper bound:

Our next result, which is similar to Theorem 4 . 4 , may be de-


duced immediately from Corollory 3.6 and Theorem 4 . 4 . Here we
offer an alternative proof which uses Propositions 3 . 4 and 4 . 1 .

Theorem 4 . 5 : For any graph G with p v e r t i c e s , d + < p


y - + I,
w i t h equality i f and only ;f G = K or
P P'
-
Proof: If G = K or K , trivially d + y = p + 1. Therefore
P P -
let G have p vertices and G # K or K
P P
. From Proposition 3.4
y L p - A , hence y -
< p - 6. We claim that this latter inequal-
ity, or d 1.6 + 1, is strict. Suppose not, then
y = p - 6 = p - d + 1. Using the fact that d(G) p/y(G),
.
d 5 p/ (p-d+l) i.e d (p-d+l) 5 p, from which we deduce

Both factors in ( 4 ) are nonnegative, hence p = d or d = 1.


In the former case G = K while d = 1 implies that G has an
P -
isolated vertex v. Since G - v # Kp-l, y (G-v) < p - 1 and y (G)< p.
In this case, since 6 = 0, the inequality y c p -
- 6 is strict.
DOMINATION I N GRAPHS 255

Thus one of t h e i n e q u a l i t i e s y < p - 8 o r d -


< 8 + 1 is
s t r i c t . Hence d + y < p + 1 a s asserted.

Corollary 3.6 and Theorem 4.4 a l s o enable u s t o sharpen


Theorem 3.6.

- y
Corollary 4.5: For any gmph G d t h p vertices + 7 -< p + 1
w i t h equazity if and only if G = K or G = K
P P’
For any graph G , a clique of G is a dominating s e t of G,
hence 7 7
< c. I n view of t h e i n e q u a l i t y < d, we were l e d t o
i n v e s t i g a t e t h e p o s s i b i l i t y t h a t c < d for-all graphs. Further
motivation w a s provided by analogy i i t h t h e concept of p e r f e c t
graphs.
A perfect graph i s a graph G which s a t i s f i e s t h e following
condition:

(4.1) x(G’) = C(G’), f o r every induced subgraph G’ of G.

P e r f e c t graphs a l s o s a t i s f y t h e following condition:

(4.2) f3,(G’) = O(G’), f o r every induced subgraph G’ of G.

I n f a c t , Lovdsz 1161 has proved t h a t Condition ( 4 . 1 ) holds


f o r a graph G i f and only i f Condition ( 4 . 2 ) holds. A very com-
p l e t e discussion of a number of i n t e r e s t i n g aspects of p e r f e c t
graphs can be found i n Berge 13, Ch.161.
For any graph G I t h e above-mentioned four parameters s a t -
i s f y t h e following r e l a t i o n s :

For dominating s e t s i n graphs, each of these parameters


has a n a t u r a l counterpart. The r e s u l t t h a t fl,(z)
= C ( G ) has
t h e following counterpart.

Proposition 4.6: For m y graph G, i (5) = c(G).

I f we l e t t h e domatic number d(G) be t h e counterpart t o


t h e chromatic number x ( G ) , then t h e counterpart t o e ( E ) i s t h e
parameter R (GI described i n t h e following paragraph.
L e t u s say t h a t a s e t of v e r t i c e s W is loose i f no vertex
not i n W is adjacent t o every vertex i n W.
A p a r t i t i o n V1, V2, ...
, Vk of V(G) i s a loose partition
of G i f every subset Vi i s loose. Let us define 2 ( G ) t o be t h e
maximum order of a loose p a r t i t i o n of G.
256 COCKAYNE AND HEDETNIEMI

Proposition 4. 7: For any graph G, d(G) = ~(z).


Thus the following analogous situation suggests itself:

once again leading us to investigate whether the inequality


c < d is true. A quick check of all graphs having p < 5 vertices
shows that for these graphs, c (GI = d (GI. A check ofall the
graphs with 6 vertices reveals several graphs for which c(G) = 2
and d(G) = 3 , yet c(G) < d(G) for all of these graphs as well.
A lot of hard work-on this conjecture uncovered the counter-
example of Figure 1 in which c(G) = 4 , yet d(G) = 3 .

Fig. 1 A counterexample to the conjecture


< d(G).
c(G) -

Attempts to settle the c < d - conjecture did reveal, how-


ever, a rather broad class of graphs for which the size of a
smallest clique is a lower bound for the domatic number.
A graph G is said to be indominable if it has a D-partition
in which every subset is independent. As such the dominating
partition is also a coloring of the graph. Note that id(G) is
positive if and only if G is indominable.

Proposition 4.8: I f G is indominable then c(G) -


< d(G).

Proof: The following sequence of inequalities obviously holds


for indominable graphs:

c(G) - < x(G) 5 id(G)


< C(G) - d(G).

Proposition 4.9: If the complement 5 of a graph G is indomin-


able, then c(G) -
< d(G).
DOMINATION IN GRAPHS 257

Proof: Let D1, ..., Dk be a partition of V(G) into sets which


are independent dominating sets of E. Since each D. is a clique
1
of G, c (G) 5 min IDi 1.
Any set, containing a vertex from each
i
Di, is a dominating set of G. There are minlDil disjoint sets
i
with that property. Hence d (GI minlDi 1 and c (G) 5 d(G) as
required. i

- Note that there do exist graphs G such that neither G nor


G is indominable, such an example is the 5-cycle,
c5
.
Attempts to produce a D-partition of order c(G) for an
arbitrary graph G, naturally lead one to consider the clique
structure of G, in particular, the clique graph K(G) of G.
When the clique structure of G is not terribly complicated,
one can often produce D-partitions of order c(G), as the next
two results indicate.

Theorem 4.10: I f K ( G ) = C where n is even, then


n
c(G) -
< d(G).

Proof: In this and the following proofs we abbreviate c(G) to


c. Let the cliques of G in order round the cycle of K(G) be
x1 ..., Xn where lXil 1. c, i = 1, ...,
n. We shall iteratively
construct a partition of V(G) into c classes 1, c. Vertex ...,
v is said to be colored if some class contains v and clique X is
said to be coZored if each class contains a vertex of X. After
the ith stage of the iteration each clique Xj' 1 -
< j -
< i will
have been colored.
Choose any set Y1 of c vertices from X 1' Let Z1 = 1'
n'
and 2 = Y1 f l X2. Then 0 -
< lZ,l, IZ21 5 c and the structure of
2
K(G) implies Z1 n Z2 = 9. Assign one element of Y to each of
1
..
the classes 1, . , c such that Z1, Z2 are assigned to the lZ1 I
lowest and IZ [ highest classes respectively. These classes
2
form the first iterate and X1 is colored.
1
Next select any set Y2 of c - lZ2 uncolored vertices of
x2 and set 2 = Y2 n X We note that 0 -
< lZ,I -
< c. Assign
3 3'
258 COCKAYNE AND HEDETNIEMI

one element of Y2 t o each of the 1Y


2
I lowest c l a s s e s so t h a t
2 occupies the lZ31 lowest classes. The c l a s s e s now form the
3
second i t e r a t e and thus f a r X1, X2 a r e colored.
I n general a t the i t h stage, i = 2 , ..., n-1, we s e l e c t
any set Yi of c - I
12. uncolored v e r t i c e s of X and s e t
1 i
= Y . n Xi+l and note t h a t 0 - < IZi+ll 5 c. The c r u c i a l
'i+l 1
f a c t i n the construction i s t h a t the assignment of Y t o the
i
classes depends on the p a r i t y of i. I n f a c t , f o r i odd (even),
1 1
1
the v e r t i c e s of Y . a r e assigned t o the IY. highest (lowest)
classes so t h a t the v e r t i c e s of Z
i+l
occupy the 12
i+l
Ihighest
(lowest) classes. The c l a s s e s then form the i t h i t e r a t e and
x1
..., X . have been colored.
1
After the (n-1Ist stage, t h e v e r t i c e s of X colored so f a r
n
are i n Z o r 2
1
. n
Since the s t r u c t u r e of K ( G ) implies

'n- 1
n X 1 = $I, w e deduce Z 1n Zn = $I. Because n i s even, the
v e r t i c e s of Z a r e i n the lZ
n
I n
highest c l a s s e s while Z1 occupies
the IZ 1 1
lowest c l a s s e s . Therefore, i f Iz,~ + lznl 2 c , xn i s
already colored, while, if not, w e may s e l e c t c - lZ,l -
IZ,I
uncolored v e r t i c e s from X and assign these t o c l a s s e s which so
n
f a r contain no vertex from X . n
Thus X may be colored.
n
Any vertex uncolored a f t e r the nth stage i s now assigned t o
the c t h c l a s s . The c l a s s e s then form a D-partition of G and
c(G) - < d(G) a s required.

Theorem 4.11: If the clique graph of G is a tree, then


c(G) - d(G).

Proof: We use the i n t e r p r e t a t i o n of the word "colored" given


i n the proof of Theorem 4.10. Suppose K(G) = T, a t r e e . W e
color the cliques of any path P i n T with c colors using the
algorithm of Theorem 4.10. We note t h a t t h i s algorithm ensures
t h a t the cliques a r e colored i n the order of t h e i r occurrence
on P and t h a t a t each stage a t m o s t one vertex i s added t o any
color c l a s s .
If any clique i s uncolored, there i s an uncolored clique X
such t h a t [ X , Y l E E ( T ) and Y i s colored. Some v e r t i c e s of X may
already be colored. These a r e necessarily i n X fl Y , since t h e
DOMINATION I N GRAPHS 259

s t r u c t u r e of K ( G ) prevents X from i n t e r s e c t i n g any other colored


clique. No two v e r t i c e s of X f7 Y may be colored t h e same, f o r
i f so, they w e r e added a t d i f f e r e n t stages i n t h e coloring of
the cliques of P. This would imply t h a t one of these v e r t i c e s
is i n another clique 2, i . e . i s i n three cliques, contrary t o
the s t r u c t u r e of K ( G ) . I t follows t h a t there are s u f f i c i e n t
v e r t i c e s i n X t o be assigned one each t o the color c l a s s e s not
y e t assigned a vertex of X and so X may be colored.
The argument of the l a s t paragraph i s now repeated with
the subtree of T, induced by X together with the v e r t i c e s of P,
replacing P. Thus we can successively color a l l cliques of G
w i t h c classes. Having done t h i s , a l l v e r t i c e s so f a r uncolored
may be colored a r b i t r a r i l y and the r e s u l t i n g color c l a s s e s form
a D-partition of G. Therefore c(G) - < d(G) a s required.

To d a t e we have been able t o show t h a t under several con-


d i t i o n s on t h e cliques of G , i f t h e clique graph of a graph i s
a cycle of odd length, then c(G) 5 d(G). W e a r e led therefore
t o o f f e r t h e following two conjectures.
Conjecture I : I f K ( G ) = Czn+l, then c ( G ) d(G).

Conjecture 2: I f K ( G ) is bipartite, then c ( G ) 2 d ( G ) .

5. SUMMARY AND OPEN PROBLEMS

The theory of domination i s i n a sense, "dual" t o the


theory of coloring. F i r s t , coloring i s based on independence,
a property which i s hereditary (cf. [91). A property P i s he-
reditary i f whenever a subset U has property P, then every sub-
set U' of U a l s o has property P. Second, f o r coloring, the
chromatic number i s the m i n i m order of an independent p a r t i -
t i o n . Domination, however, i s an expanding property, (cf. [91)
i n t h a t any superset of a dominating s e t i s a l s o a dominating
s e t ; furthermore, the domatic number i s the rnax+mun order of a
D-partition of G.
A d i s s i m i l a r i t y i s the following. Independence can be s a i d
t o be a "local" property i n t h a t i n order t o determine whether
o r not a s e t of v e r t i c e s W i s independent, one need only consider
the subgraph <W> induced by the s e t W. Dominance, on the other
hand, could be s a i d t o be a "global" property i n t h a t i n order t o
determine whether o r not a set of v e r t i c e s W i s a dominating s e t ,
one has t o consider the r e l a t i o n s h i p between W and the e n t i r e
graph G.
We have attempted i n t h i s paper t o i n i t i a t e the development
of a theory of domination.
260 COCKAYNE AND HEDETNIEMI

Further topics for study are properties of indominable


graphs, relationships among a variety of parameters of a graph
relating to dominating sets and the difficult problem of con-
structing a fast algorithm for finding the domination number of
a graph.
A large number of problems are open for study for the do-
matic number; among which we find the following to be particu-
larly worthwhile :

1. Characterize the class of full graphs. The computa-


tion of d(G) is a difficult problem, in general. However, for
full graphs, the computation of d = 6 + 1 is trivial.
2. Characterize the class of graphs whose domatic number
is 2.
3. Characterize the class of graphs for which c I d .

6. ACKNOWLEDGEMENTS

Dr. Cockayne wishes to thank the Canadian National Research


Council for the support of Research Grant NRC A7544.
Dr. Hedetniemi wishes to thank the University of Victoria,
B. C., Canada, for its kind hospitality during the summer of
1974.
The authors gratefully acknowledge the recommendations of
the referees, which did much to improve the paper.

7. REFERENCES

1. Aigner, M., "Some Theorems on Coverings," Studia Sci. Math.


HwZgar., 5 , 1970, pp. 303-315.

2. Berge, C., Theozy of Graphs and i t s Applications, Methuen,


London, 1962.

3. Berge, C. , Graphs and Hypergraphs, North-Holland, Amsterdara,


1973.

4. Billera, L. J., "On the Composition and Decomposition of


Clutters," J . Comb. Theory, 1 1 ~ ,1971, pp. 243-245.

5. Chartrand, G. and D. Geller, "Uniquely Colorable Planar


Graphs," J . Comb. Theory, 6, 1969, pp. 271-278.

6. Chartrand, G . , D. Geller and S. Hedetniemi, "Graphs with


Forbidden Subgraphs,It J . Comb. Theory, 10, 1971, pp. 12-41.
DOMINATION I N GRAPHS 261

7. Cockayne, E. J. and S. T. Hedetniemi, "Independence Graphs,"


Proc. of 5th S.E. Conference on Combimtorics, Graph Theory
and Computing, Boca Raton, F l o r i d a , 1974.

8. Cockayne, E. J . a n d S. T.Hedetniemi, " O p t i m a l Domination i n


Graphs," IEEETrans. Circuits &Systems, CAS-22, 1975, pp. 41-44.

9. Cockayne, E. J. and S. T. Hedetniemi, " I n t e r p o l a t i o n Sys-


t e m s , " Proc. of 3rd S.E. Conference on Combinatorics, Graph
Theory and Computing, 1972, pp. 117-130.
10. Edmonds, J. and D. R. Fulkerson, "Bottleneck Extrema, " J.
Comb. Theory, 8, 1970, pp. 299-306.
11. Gupta, R. P . , "Independence and Covering Numbers of Line
Graphs and Total Graphs," Proof Techniques in Graph Theory,
F. Harary, ed., Academic P r e s s , New York, 1969.

12. Haraw, F. , Graph Theory, Addison-Wesley, Reading, Massa-


c h u s e t t s , 1969.

13. J a e g a r , F. and C. Payan, "Relations du Type Nordhaus-Gaddum


Pour l e Nombre d'Absorption d'un Graphe Simple," C. R. Acad.
Sc. P d s , S e r i e s A, t 274, 1972, pp. 728-730.

14. Kalbfleisch, J. G., R. G. Stanton and J. D. Horton, "On


Covering S e t s and Error-Correcting Codes," J . Comb. Theory,
1 1 A , 1971, pp. 233-250.

15. L i u , c. I,., Introduction t o CombinutoriaZ Mathematics,


McGraw-Hill, New York, 1968.

16. L O V ~ S Z ,L., "Normal Hypergraphs and t h e P e r f e c t Graph Con-


j e c t u r e , * ' Discrete Math, 2 , 1972, pp. 253-267.

17. Nieminen, J . , "Two Bounds f o r t h e Domination Number of a


Graph," J . I n s t . Maths. A p p l i e s . , 1 4 , 1974, pp. 183-187.

18. Nordhaus, E. A. and J. W. Gaddum, "On Complementary Graphs,"


Amer. Math. Monthly, 63, 1956, pp. 175-177.
19. O r e , O., Theory of Graphs, Amer. Math. SOC. Colloq. Publ.,
38, Providence, 1962.

20. Vizing, V. G., "A Bound on t h e External S t a b i l i t y Number of


a Graph," DokZady A . N., 164, 1965, pp. 729-731.

Paper received June 2, 1975.

You might also like