You are on page 1of 7

AKCE International Journal of Graphs and Combinatorics

ISSN: 0972-8600 (Print) 2543-3474 (Online) Journal homepage: https://www.tandfonline.com/loi/uakc20

The -distance chromatic number of trees and


cycles

Niranjan P.K. & Srinivasa Rao Kola

To cite this article: Niranjan P.K. & Srinivasa Rao Kola (2019) The -distance chromatic number of
trees and cycles, AKCE International Journal of Graphs and Combinatorics, 16:2, 230-235, DOI:
10.1016/j.akcej.2017.11.007

To link to this article: https://doi.org/10.1016/j.akcej.2017.11.007

 © 2017 Kalasalingam University.


Production and Hosting by Elsevier B.V.

Published online: 10 Jun 2020.

Submit your article to this journal

Article views: 188

View related articles

View Crossmark data

Citing articles: 1 View citing articles

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=uakc20
Available online at www.sciencedirect.com
ScienceDirect
AKCE International Journal of Graphs and Combinatorics 16 (2019) 230–235
www.elsevier.com/locate/akcej

The k-distance chromatic number of trees and cycles


Niranjan P.K. ∗, Srinivasa Rao Kola
Department of Mathematical and Computational Sciences, National Institute of Technology Karnataka, Surathkal, India

Received 18 April 2017; received in revised form 23 November 2017; accepted 24 November 2017
Available online 18 December 2017

Abstract

For any positive integer k, a k-distance coloring of a graph G is a vertex coloring of G in which no two vertices at distance
less than or equal to k receive the same color. The k-distance chromatic number of G, denoted by χk (G) is the smallest integer
α for which G has a k-distance α-coloring. In this paper, we improve the lower bound for the k-distance chromatic number of an
arbitrary graph for k odd case and see that trees achieve this lower bound by determining the k-distance chromatic number of trees.
Also, we find k-distance chromatic number of cycles and 2-distance chromatic number of a graph G in which every pair of cycles
are edge disjoint.
⃝c 2017 Kalasalingam University. Production and Hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND
license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Keywords: Distance coloring; k-distance chromatic number; 2-distance chromatic number

1. Introduction
In 1969, Kramer and Kramer [1] have defined, for any positive integer k, the k-distance coloring of a graph G as
an assignment f of positive integers to the vertices of G such that for any vertices u and v of G, | f (u) − f (v)| ⩾ 1
if d (u, v) ⩽ k. If α is the maximum integer (color) assigned by f , then f is referred as k-distance α-coloring. The
k-distance chromatic number χk (G) is the smallest α for which G has a k-distance α-coloring. Since k ⩾ diam (G)
is trivial case, we consider k < diam (G) in this paper. For a non-negative integer r and for a vertex v of a graph G,
the graph G rv denotes the subgraph of G induced by the vertices of G which are at distance less than or equal to r
from v. Sharp [2] have given a lower bound for k-distance chromatic number of an arbitrary graph as below.

Theorem 1.1 ([2]). For any graph G and positive integer k,


⎧ ⏐ ( )⏐
⏐ k ⏐
2

⎪ max
⎨v∈V (G) ⏐⏐V G v ⏐⏐ if k is even,
χk (G) ⩾ ⏐ ( )⏐
k−1 ⏐

⎩ max ⏐⏐V G v 2 ⏐⏐ + 1

⎪ if k is odd.
v∈V (G)

Peer review under responsibility of Kalasalingam University.


∗ Corresponding author.
E-mail addresses: niranjanpk704@gmail.com (Niranjan P.K.), srinu.iitkgp@gmail.com (Srinivasa Rao K.).

https://doi.org/10.1016/j.akcej.2017.11.007
0972-8600/⃝ c 2017 Kalasalingam University. Production and Hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND
license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Niranjan P.K. and Srinivasa Rao K. / AKCE International Journal of Graphs and Combinatorics 16 (2019) 230–235 231

Kramer and Kramer [1,3] while characterizing graphs with χk (G) = k + 1 proved that χk (Cn ) = k + 1 if n is a
multiple of k + 1 and χk (Pn ) = k + 1 if n ⩾ k + 1. For any planar graph G with maximum degree ∆, Wegner [4]
proved that χ2 (G) ⩽ 8, if ∆ ⩽ 3 and conjectured as below.

Conjecture 1.2 ([4]). For any planar graph G with maximum degree ∆,


⎪ 7 if ∆ = 3,
∆ + 5 if 4 ⩽ ∆ ⩽ 7,

χ2 (G) ⩽ ⌊ 3∆ ⌋
if ∆ ⩾ 8.



2
The average degree of a graph G, denoted ad (G), is |V (G)|
1
d (v). The maximum average degree of a graph

v∈V (G)
G, denoted mad (G), is the maximum of ad (H ) on every subgraph H of G. Bonamy et al. [5] have found that the
2-distance chromatic number of a graph with maximum average degree less than 73 and maximum degree ∆ ⩾ 4
as ∆ + 1. Also, in [6] they have determined that χ2 (G) is ∆ + 1 for any planar graph G with ∆ ⩾ 5 and girth
g (G) ⩾ 12; ∆ ⩾ 6 and g (G) ⩾ 10; ∆ ⩾ 8 and g (G) ⩾ 9. Bu and Lv [7] have given an upper bound for χ2 (G) of a
planar graph without cycles of length 3, 4 and 7 and ∆ ⩾ 15 as ∆ + 4. Kramer and Kramer [8] have given an upper
bound for χ3 (G) of bipartite graph G as 2 (1 + ∆ (∆ − 1)). Also, they have proved, χ3 (G) ⩽ 8 for a bipartite planar
graph G with maximum degree ∆ ⩽ 3. A survey paper on k-distance coloring of graphs is published by Kramer and
Kramer [9].
In this paper, we improve the lower bound for the k-distance chromatic number of an arbitrary graph given by
Sharp [2] when k is odd. For any k, we prove that trees attain the lower bound and hence we determine χk (T ) for any
tree T . Also, we find χk (Cn ) for any cycle Cn . Further, we determine χ2 (G) for a graph G with no two cycles share
an edge.

2. Trees
We start this section with a lower bound for k-distance chromatic number of an arbitrary graph and see that this is
an improvement of lower bound given by Sharp [2] for k odd case. Later, we prove that trees attain the lower bound.
Before we give the lower bound for χk (G), similar to G rv , we define G ruv for any edge uv of G

Definition 2.1. For any non-negative integer r and an edge uv of a graph G, the graph G ruv denotes the subgraph of
G induced by the vertices of G which are at distance less than or equal to r from u or v.

Theorem 2.2. If G is a graph and k is any positive integer, then


⎧ ⏐ ( )⏐
⏐ k ⏐
2

⎪ max
⎨v∈V (G) ⏐⏐V G v ⏐⏐ if k is even,
χk (G) ⩾ ⏐

( )⏐
k−1 ⏐
⎩ max ⏐⏐V G uv2 ⏐⏐

⎪ if k is odd.
uv∈E(G)

Proof. If k is even, the result follows from Theorem 1.1. Suppose that k is odd. Let uv be an edge and w, w′
k
be any two vertices of 2
( G uv′ ). Then d (u, w) ⩽ k−1 2 (
or d (v,) w) ⩽ 2 (. Without
k−1
loss of generality, we assume that
d (u, w) k−1
w k−1
w, w v, w k−1
′ ′
)
⩽ 2
. If d u, ⩽ 2
, then clearly d ⩽ k. If d ⩽ 2
, then it is easy to observe that
d w, w ⩽ k as u and v⏐ are(adjacent. w
( ′
)
)⏐ Therefore, in any k-distance coloring, u and should receive different colors.
k−1 ⏐
Hence χk (G) ⩾ max ⏐⏐V G uv2 ⏐⏐. □

uv∈E(G)

In a graph G, eG (u) (or simply e (u)) denotes the eccentricity of a vertex u in G, diam (G) and rad (G) are diameter
and radius of G respectively.

Lemma 2.3. Let T be a tree with n vertices. Let Ti−1 = Ti − vi , i =( n,)n − 1, . . . , 2, 1, where Tn = T and vi is a
vertex of Ti such that eTi (vi ) = diam (Ti ). If T ′ = (Ti )rvi , then diam T ′ ⩽ r .
232 Niranjan P.K. and Srinivasa Rao K. / AKCE International Journal of Graphs and Combinatorics 16 (2019) 230–235

Proof. On contrary, suppose diam T( ′ > )r . Since eTi( (vi)) = diam (Ti ), vi is a leaf of Ti and so in T ′ . Let v j and
( )

vl be antipodal vertices of T ′ (i.e. d v j , vl = diam T ′ ). Let P j and Pl be the v j − vi and vl − vi paths in T ′


respectively. Since vi is a leaf of T ′ and the paths P j and Pl end at vi , there must be a vertex common to them other
than vi . Let u be the first vertex of P j which is in Pl . Then ( vj − ) u subpath ( u −) vl subpath of Pl
)of P j followed by
give the v j − vl path in T ′ . If d (u, vi ) ⩾ d (u, vl ), then d v j , vl = d v j , u + d (u, vl ) ⩽ d v j , u + d (u, vi ) ⩽ r ,
(

a contradiction. So d (u, vi ) < d (u, vl ).


Using the above inequality d (u, vi ) < d (u, vl ), we show that there is a path in Ti whose length is greater
than the diam (Ti ) which is a contradiction. Let v be a vertex of Ti such that d (v, vi ) = diam (Ti ). Let P be
the vi − v path in Ti . Since vi is a leaf and the paths P j and P end at vi , there must be a vertex common to
them other than vi . Let w be the first vertex of P which is also in P j . If w is in v j − u subpath of P j , then
v − w subpath of P followed by w − u subpath of P j , followed by u − vl subpath of Pl is the v − vl path. ( So
d (v, vl ) = d (v, w) + d (w, u) + d (u, vl ) = d (v, u) + d (u, vl ) > d (v, u) + d (u, vi ) = d (v, vi ) = diam T ′ ,
)

a contradiction. Therefore w cannot be in v j − u subpath of P j . If w is in u − vi subpath of P j , then v − w


subpath of P followed by w − u subpath of P j , followed by u − vl subpath of Pl is the v − vl path. ( )So
d (v, vl ) = d (v, w) + d (w, u)(+ )d (u, vl ) = d (v, u) + d (u, vl ) > d (v, u) + d (u, vi ) ⩾ d (v, vi ) = diam T ′ , a
contradiction. Therefore diam T ′ ⩽ r . □
Griggs and Yeh [10] have introduced the concept of L (2, 1)-coloring as a variation of channel assignment problem.
An L (2, 1)-coloring of a graph is a coloring of the vertices with non-negative integers such that the colors on adjacent
vertices differ by at least 2 and the colors on vertices at distance 2 differ by at least 1. In the introductory paper, they
have proved that the L (2, 1) chromatic number of tree with maximum degree ∆ is either ∆ + 1 or ∆ + 2 by giving
an L (2, 1)-coloring. Motivated by this, we give a k-distance coloring of a tree in the theorem below.

Theorem 2.4. For any tree T ,


⎧ ⏐ ( )⏐
⏐ k ⏐

⎪ max ⏐⏐V Tv2 ⏐⏐
⎨v∈V if k is even,
(G)
χk (T ) = ⏐ (

)⏐
k−1 ⏐
2

⎩ max ⏐V Tuv ⏐⏐
⎪ ⏐ if k is odd.
uv∈E(G)

Proof. From Theorem 2.2 it is clear that


⎧ ⏐ ( )⏐
⏐ k ⏐

⎪ max ⏐⏐V Tv2 ⏐⏐
⎨v∈V if k is even,
(G)
χk (T ) ⩾ ⏐ ( )⏐
k−1 ⏐

⎩ max ⏐⏐V Tuv2 ⏐⏐

⎪ if k is odd.
uv∈E(G)
⏐ ( )⏐
k ⏐
Now, we give a k-distance α-coloring for the tree T , where α = max ⏐⏐V Tv2 ⏐⏐, if k is even and α =

⏐ ( )⏐ v∈V (G)
k−1 ⏐
max ⏐⏐V Tuv ⏐⏐, if k is odd. Let Ti and vi , i = 1, 2, 3, . . . , n are subtrees and vertices of T respectively as
⏐ 2
uv∈E(G)
given in Lemma 2.3. We assign the color 1 to v1 and( 2 )to v2 . Suppose v1 , v2 , . . . , vi−1 are colored. Now, to color vi ,
consider the tree T ′ = (Ti )kvi . By Lemma 2.3, diam T ′ ⩽ k.

Case I: k is even.
( ) ⌈ diam (T ′ ) ⌉ ⌈ k ⌉ k
Let v be a vertex of T ′ such that eT ′ (v) = rad T ′ = 2
⩽ 2 = 2 . Then every vertex of T ′ is distance at
⏐ ( )⏐ ⏐ ( )⏐
k k ⏐ k ⏐
most 2 from v. So T is a subgraph of Tv . Since V T − {v} < V T ⩽ ⏐V Tv ⏐ ⩽ max ⏐⏐V Tv2 ⏐⏐ = α,
k ′ 2
⏐ ( ′) ⏐ ⏐ ( ′ )⏐ ⏐ 2 ⏐

⏐ ⏐ ⏐ ⏐ ⏐
v∈V (G)
we have at least on color not used in T ′ to color vi .

Case II: k is odd.


( )
Subcase (i): diam T ′ is odd.
Niranjan P.K. and Srinivasa Rao K. / AKCE International Journal of Graphs and Combinatorics 16 (2019) 230–235 233

Fig. 1. A 4-distance coloring of a tree as in Theorem 2.4.

diam (T ′ )
⌈ ⌉
Since diameter of T ′ is odd, the center of T ′ is an edge, say uv. So eT ′ (u) = eT ′ (v) = rad T ′ =
( )
2

v. w
⌈k ⌉ k+1 ′ k+1 ′
2
= 2
. Therefore, every vertex of T is distance at most 2
from u and If is a vertex of T with
d (w, u) = k+1 2
, then d (w, v) ⩽ k−1
2
and vice versa. So every vertex of T ′ is⏐ at(distance )⏐ less than or⏐equal
( to k−12
)⏐ from
k−1 k−1 ⏐ k−1 ⏐
u or v. So T ′ is a subgraph of Tuv2 . Since ⏐V T ′ − {v}⏐ < ⏐V T ′ ⏐ ⩽ ⏐⏐V Tuv2 ⏐⏐ ⩽ max ⏐⏐V Tuv2 ⏐⏐ = α,
⏐ ( ) ⏐ ⏐ ( )⏐ ⏐ ⏐
uv∈E(G)
we have a color not used in T ′ to color vi .
( )
Subcase (ii): diam T ′ is even.
diam (T ′ )
⌈ ⌉
Let v be the center of T ′ . Then eT ′ (v) = rad T ′ = ⩽ k−1 = k−1
( ) ⌈ ⌉
2 2
. Therefore, every vertex of T ′
2 ⏐ ( )⏐ ⏐ ( )⏐
k−1 ⏐ k−1 ⏐
v. ⏐V T − {v}⏐ < ⏐V T ′ ⏐ ⩽ ⏐V Tv 2 ⏐ ⩽ ⏐V Tuv2 ⏐ ⩽
⏐ ( ′) ⏐ ⏐ ( )⏐ ⏐ ⏐
is at distance less than or equal to k−1 2
from Since ⏐ ⏐ ⏐ ⏐
⏐ ( )⏐
k−1 ⏐
max ⏐⏐V Tuv2 ⏐⏐ = α, where u is any neighbor of v in T ′ , we have a color not used in T ′ to color vi . □

uv∈E(G)

Example 2.5. Fig. 1 is a 4-distance coloring of a tree T with 42 vertices as in Theorem 2.4. A 3-distance coloring of
the same tree is in Fig. 2.

3. Cycles
Kramer and Kramer [1,3] determine that χk Cl(k+1) is k + 1. Now, we find the k-distance chromatic number for
( )
any cycle Cn .

Theorem 3.1. For any cycle Cn , χk (Cn ) = k + 1 + , where r and l are integers such that n = l (k + 1) + r ,
⌈r ⌉
l
0 ⩽ r < k + 1.

Proof. In any k-distance coloring of a graph G if two vertices u and v receive same color, then d (u, v) is at least
k + 1. Since n = l (k + 1) + r , any k-distance
⌈ ⌉coloring of Cn⌈ assign a color c to at most l vertices of Cn . Therefore,
any k-distance coloring of Cn needs at least nl = k + 1 + rl colors. Now, we show that χk (C )
⌉ ⌈r ⌉
n ⩽ k + 1 + l
by defining a k-distance coloring of Cn . Let Cn : v1 , v2 , . . . , vn , v1 be the cycle and α = k + 1 + rl .
⌈ ⌉
Now,
n = l (k + 1) + r
⌈r ⌉
= l (k + 1) + l ′ + r′
l
234 Niranjan P.K. and Srinivasa Rao K. / AKCE International Journal of Graphs and Combinatorics 16 (2019) 230–235

Fig. 2. A 3-distance coloring of a tree as in Theorem 2.4.

⌈r ⌉
where, r ′ and l ′ are integers such that 0 ⩽ r ′ <
( ⌈ r ⌉) ( l
= l′ k + 1 + + l − l ′ (k + 1) + r ′
)
)l
= l ′ α + l − l ′ (k + 1) + r ′ .
(

It is easy to see that l ′ ⩽ l. Define a map f from {v1 , v2 , . . . , vn } to {1, 2, . . . , α} by


f (vi ) = t if i ≡ t mod α, 1 ⩽ i ⩽ l ′ α.
f vl ′ α+ j = t if j ≡ t mod (k + 1) , 1 ⩽ j ⩽ l − l ′ (k + 1) = n − r ′ .
( ) ( )

f (vn−s ) = α − s, 0 ⩽ s < r ′ .
It is easy to verify k-distance coloring condition for vi and v j , 1 ⩽ i, j ⩽ n − r ′ . Since the color given to vn−(r ′ −1)
is α − r ′ − 1 > α − rl + 1 ⩾ k + 2 > k + 1 the k-distance coloring condition is satisfied between vi and vn−s ,
( ) ⌈ ⌉
1 ⩽ i ⩽ l ′ α, 0 ⩽ s < r ′ ; and vl ′ α+ j and vn−s , 1 ⩽ j ⩽ n − r ′ , 0 ⩽ s < r ′ . □

4. 2-distance chromatic number of a class of graphs


In this section, we determine 2-distance chromatic number of a graph in which every pair of cycles is edge disjoint.

Theorem 4.1. Let G be a graph with maximum degree ∆ ⩾ 3. If every pair of cycles in G is edge disjoint, then
{
5 if G contains C5 and ∆ = 3,
χ2 (G) =
∆+1 otherwise.

Proof. Let G be a graph with maximum degree ∆ ⩾ 3. Let α = 5 if G contains C5 and ∆ = 3, otherwise α = ∆ + 1.
Since χ2 (C5 ) = 5 and from Theorem 2.2 we have χ2 (G) ⩾ α. Now, we give a procedure to define a 2-distance
coloring of G using α colors. Let C be any cycle in G. It is clear that χ2 (C) ⩽ α. We color the vertices of C using
Theorem 3.1. Let u be a vertex on C with degree of u, deg (u) ⩾ 3. As deg (u) ⩽ ∆, u has at most ∆ − 2 neighbors
which are not colored. Since α ⩾ ∆+1, we have at least ∆−2 colors (excluding the colors given to u and its neighbors
) C . Let C : uu 3 u 4 . . . u m u 1 u and c1 , c2
′ ′
on C) to color the neighbors of u not on C. Suppose that u is on any other(cycle
and c3 be the colors assigned to u 1 , u and u 3 respectively. We choose χ2 C − 3 number of colors c4 , c5 , . . . , cχ2 (C ′ )

from {1, 2, 3, . . . , α} other than c1 , c2 , c3 . We color the vertices of C ′ using colors c1 , c2 , . . . , cχ2 (C ′ ) as in Theorem 3.1
(color ci refers to color i in Theorem 3.1). Now, we choose a colored vertex of G which has uncolored neighbors and
continue as above until all the vertices of G are colored. It is easy to see that α is maximum color and it is used to
either a vertex in C5 or to a maximum degree vertex or any one of its neighbors. □
Niranjan P.K. and Srinivasa Rao K. / AKCE International Journal of Graphs and Combinatorics 16 (2019) 230–235 235

5. Conclusion
In this article, k-distance chromatic number of trees, cycles and 2-distance chromatic number of a graph with no
two cycles share an edge are determined. It is clear that all the above graphs satisfy Conjecture 1.2.
From Theorem 2.4 and from Theorem 4.1, it is easy to see that the 2-distance chromatic number of a graph G with
no two cycles share an edge is at most one more than the 2-distance chromatic number of a tree T obtained from G by
deleting exactly one edge from each cycle in G. The procedure given in Theorem 4.1 becomes difficult as k increases.
Since any unicyclic graph can be obtained by an edge to a tree, one can try to find the k-distance chromatic number
of unicyclic graph.

References
[1] Florica Kramer, Horst Kramer, Un probleme de coloration des sommets d’un graphe, C. R. Acad. Sci. Paris A 268 (7) (1969) 44–46.
[2] Alexa Sharp, Distance coloring, in: European Symposium on Algorithms, Springer, Berlin Heidelberg, 2007, pp. 510–521.
[3] Florica Kramer, Horst Kramer, Ein Färbungsproblem der Knotenpunkte eines Graphen bezüglich der Distanz p, Rev. Roumaine Math. Pures
Appl. 14 (2) (1969) 1031–1038.
[4] G. Wegner, Graphs with given diameter and a coloring problem, Technical Report, University of Dortmund, 1977.
[5] Marthe Bonamy, Benjamin Lévêque, Alexandre Pinlou, 2-distance coloring of sparse graphs, J. Graph Theory 77 (3) (2014) 190–218.
[6] Marthe Bonamy, Benjamin Lévêque, Alexandre Pinlou, 2-distance coloring of sparse graphs, in: Eurocomb’11: European Conference on
Combinatorics, Graph Theory and Applications, in: Electronic Notes in Discrete Mathematics, vol. 38, 2011, pp. 155–160.
[7] Yuehua Bu, Xia Lv, 2-Distance coloring of a planar graph without 3, 4, 7-cycles, J. Comb. Optim. 32 (2016) 244–259.
[8] Florica Kramer, Horst Kramer, On the generalised chromatic number, Ann. Discrete Math. 30 (1986) 275–284.
[9] Florica Kramer, Horst Kramer, A survey on the distance-coloring of graphs, Discrete Math. 308 (2) (2008) 422–426.
[10] Jerrold R. Griggs, Roger K. Yeh, Labelling graphs with a condition at distance 2, SIAM J. Discrete Math. 5 (4) (1992) 586–595.

You might also like