0% found this document useful (0 votes)
15 views24 pages

Approximation Algorithm For Minimum Q-Dominator Partization Problem

A dominator coloring of a graph G is a proper vertex coloring with a domination property that each vertex dominates a color class. The minimum number of colors used in a dominator coloring of G is called dominator chromatic number of G. Graphs with dominator chromatic number less than 3 have characterizations and such graphs can be recognized in polynomial time. In this paper, we investigate the computational complexity of the corresponding node deletion problem.

Uploaded by

Sayani Das
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
15 views24 pages

Approximation Algorithm For Minimum Q-Dominator Partization Problem

A dominator coloring of a graph G is a proper vertex coloring with a domination property that each vertex dominates a color class. The minimum number of colors used in a dominator coloring of G is called dominator chromatic number of G. Graphs with dominator chromatic number less than 3 have characterizations and such graphs can be recognized in polynomial time. In this paper, we investigate the computational complexity of the corresponding node deletion problem.

Uploaded by

Sayani Das
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 24

July 31, 2023 11:18 WSPC/S1793-8309 257-DMAA 2250188

Discrete Mathematics, Algorithms and Applications


Vol. 16, No. 1 (2024) 2250188 (24 pages)

c World Scientific Publishing Company
DOI: 10.1142/S1793830922501889

Approximation algorithm for minimum q-dominator


partization problem

Sayani Das
Theoretical Computer Sciences
The Institute of Mathematical Sciences
Chennai 600113, India
[email protected]

Sounaka Mishra∗
Department of Mathematics
IIT Madras, Chennai 600036, India
[email protected]

Received 22 October 2022


Accepted 19 November 2022
Published 9 January 2023

Communicated by Dachuan Xu

A dominator coloring of a graph G is a proper vertex coloring with a domination prop-


erty that each vertex dominates a color class. The minimum number of colors used in a
dominator coloring of G is called dominator chromatic number of G and is denoted as
χd (G). Graphs with χd (G) ≤ 3 have characterizations and such graphs can be recog-
nized in polynomial time. However, for k ≥ 4, deciding whether χd (G) ≤ k is NP-hard.
In this paper, we investigate the computational complexity of minimum q-dominator
partization problem (Min-q-D-Partz). In Min-q-D-Partz, given a graph G, the objective
is to find a vertex set S of minimum size such that χd (G[V \S]) ≤ q. We prove that
Min-2-D-Partz is APX-complete and has a two-factor approximation algorithm. We also
prove that Min-3-D-Partz cannot be approximated
√ within any constant factor and can
be approximated within a factor of O( log n).

Keywords: Node deletion; 2-dominator partization; 3-dominator partization; APX-hard;


approximation algorithm.

Mathematics Subject Classification 2020: 05C15, 05C69, 05C85, 68W25

1. Introduction
We consider a simple undirected graph G = (V, E) with vertex set V and edge set
E. Throughout this paper, we assume that |V | = n and |E| = m. A subset I ⊆ V

∗ Corresponding author.

2250188-1
July 31, 2023 11:18 WSPC/S1793-8309 257-DMAA 2250188

S. Das & S. Mishra

is an independent set in G if G[I] has no edge. A proper vertex coloring of a graph


G is a vertex coloring such that adjacent vertices do not have same color and the
set of vertices having same color is called a color class. Given a graph G = (V, E),
in minimum chromatic number problem it is required to compute a proper vertex
coloring of G using least number of colors. The minimum number of colors used in
a proper vertex coloring of G is called the chromatic number of G and it is denoted
as χ(G). A dominating set in G is a vertex set D such that D ∩ N [v] = ∅, for each
vertex v ∈ V . In minimum dominating set problem, it is required to compute a
dominating set D of minimum size in a given graph G and the size of a minimum
dominating set is denoted by γ(G). These two problems are known NP-complete
problems. Unless P = NP, minimum chromatic number cannot be approximated
within a factor of n1−ε , for any ε > 0 [26]. Whereas γ(G) can be approximated
within a factor of O(log n) [6] and it is the best possible approximation factor [10].
A subset S of V is dominated by a vertex v in G if S ⊆ N [v]. If S is an indepen-
dent set in G with at least two vertices and is dominated by a vertex v then v ∈ / S.
Different types of vertex coloring problems have been defined by imposing domi-
nation property on the color classes of a proper vertex coloring, such as dominator
coloring [14], domination coloring [25] and dominated coloring [21, 23, 24]. Domi-
nated coloring is also known as class domination coloring (in short cd-coloring). In
these problems, aim is to partition the vertex set V into independent sets (or color
classes) V1 , V2 , . . . , Vk along with domination property. If each vertex v ∈ V dom-
inates some color class then this k partition of V is called a k-dominator coloring
of G and the dominator chromatic number of G, χd (G), is the minimum number of
colors used in a dominator coloring of G. A k partition of V is called a k-domination
coloring if each color class is dominated by a vertex and each vertex is dominat-
ing a color class. The domination chromatic number of G, χdd (G), is the minimum
number of colors used in a domination coloring of G. If each color class of G is
dominated by a vertex then the k partition of G is called a k-cd-coloring of G.
The cd-chromatic number of G, χcd (G), is the minimum number of colors used in a
cd-coloring of G.
Minimum domination coloring was introduced by Zhou and Zhao [25]. They
proved that it is NP-complete by establishing a reduction from minimum chromatic
number. They also present some results on graphs with χdd (G) = χ(G). Min-
imum cd-coloring was introduced by Swaminathan and Sundareswaran [24] and
they studied χcd (G) for some subclasses of graphs. This problem is also known as
minimum dominated coloring problem [21]. Merouane et al. [21] established rela-
tionship between cd-coloring and total domination number. They also proved that
minimum cd-coloring is NP-complete for graphs with χcd (G) ≥ 4 and designed
polynomial time algorithm for recognizing graphs with χcd (G) ≤ 3. It is NP-hard
1−ε
to approximate minimum cd-coloring within a factor of n 1.5+1 , for any ε > 0, and
for bipartite graphs it is approximated within a factor of O(log n) [4]. Shalu et al.
[23] gave a lower bound on χcd (G) using maximum subclique problem and proved
that minimum cd-coloring is NP-complete for chordal graphs. In [8] authors proved

2250188-2
July 31, 2023 11:18 WSPC/S1793-8309 257-DMAA 2250188

Approximation algorithm for minimum q-dominator partization problem

that minimum cd-coloring is hard to approximate within a factor of ( 12 − ε) ln n, for


any ε > 0, when restricted to bipartite graphs.
In [19], authors have defined a node deletion problem associated with minimum
cd-coloring problem called minimum q-cd-partization problem. In this problem,
given a graph G = (V, E) and a positive integer q, aim is to find a vertex set S of
minimum size such that χcd (G[V \S]) ≤ q. They proved that it is NP-complete for
q ∈ {2, 3} and, for q ≥ 4, this problem is not in NP. Recently, it is being proved that
minimum 2-cd partization is APX-complete and is approximable within a factor of
2, and minimum 3-cd partization is APX-hard and can be approximable within a

factor of O( log n) [7].
Gera et al. [14] introduced the minimum dominator coloring problem and proved
that it is NP-complete by a reduction from minimum chromatic number. They also
studied this problem for different graph classes and compared χd (G) with χ(G)
and γ(G) [12–14]. For P4 -free graphs χd (G) can be computed in polynomial time,
also given a graph G it can be decided in polynomial time whether χd (G) ≤ 3 [3].
It is known to be NP-complete for bipartite, split or planar graphs and it has an
FPT algorithm for chordal graphs [2]. In [13], Gera proved that for bipartite graphs
γ(G) ≤ χd (G) ≤ γ(G) + 2. From its proof it follows that, for bipartite graphs,
minimum dominator coloring can be approximated within a factor of O(log n) as
minimum dominating set is approximable with a factor of O(log n). In [8], the
authors proved that minimum dominator coloring problem is NP-hard to approxi-
mate within a factor of n1−ε , for any ε > 0.
Like minimum q-cd partization problems, here we consider minimum q-
dominator partization problems for q ∈ {2, 3}. Given a graph G = (V, E)
and a positive integer q, in minimum q-dominator partization problem (in short
Min-q-D-Partz) it is required to find a vertex set S of minimum size such that
χd (G[V \S]) ≤ q. A graph G is q-dominator colorable if χd (G) ≤ q. Given a graph
G = (V, E), a vertex set S ⊆ V is called a q-dominator partization set in G if
χd (G[V \S]) ≤ q.
In this paper, we prove that Min-2-D-Partz is APX-complete. We also gave the
best constant factor approximation algorithm for it. We prove that Min-3-D-Partz

is APX-hard and can be approximable within a factor of O( log n). Our algorithm
is based on a characterization of graphs with χd (G) ≤ 3 available in [3, 12].

2. Minimum 2-Dominator-Partization
Here, we will consider the approximability of Min-2-D-Partz based on the charac-
terization of 2-dominator colorable graphs given in Proposition 2.1.
Proposition 2.1 ([3, 12]). If G = (V, E) is a graph with χd (G) ≤ 2, then G
satisfies one of the following propertie:

(i) G is isomorphic to K1 ,
(ii) G is isomorphic to 2K1 ,

2250188-3
July 31, 2023 11:18 WSPC/S1793-8309 257-DMAA 2250188

S. Das & S. Mishra

(iii) G is isomorphic to Kr,s with r ≥ 1 and s ≥ 1.


Let G1 = (V1 , E1 ) and G2 = (V2 , E2 ) be two graphs with V1 ∩ V2 = ∅. G1 ∪ G2 =
(V1 ∪ V2 , E1 ∪ E2 ), is the union of G1 and G2 . G1 ⊕ G2 = (V1 ∪ V2 , E1 ∪ E2 ∪ E1,2 )
is the join of G1 and G2 , where E1,2 = {(u, v) | u ∈ V1 , v ∈ V2 }. A biclique is a
complete bipartite graph Kr,s with r ≥ 1 and s ≥ 1. If G = (V, E) is a biclique
then the vertex set V has a unique bipartization V = X ∪˙ Y such that all the
edges of G are across the vertex sets X and Y . A subset S ⊆ V is a vertex
cover in G if S ∩ {u, v} = ∅, for all (u, v) ∈ E. In minimum vertex cover problem
(Min-VC) it is required to find a vertex cover S of minimum size in a given simple
graph G.
Proposition 2.2 ([11]). Min-VC is approximated within a factor of 2.
In [9], the authors have proved that minimum 2-domination-partization problem
is APX-hard by establishing an L-reduction [20] from minimum vertex cover prob-
lem. We observe that the same reduction is an L-reduction from minimum vertex
cover problem to Min-2-D-Partz. Therefore, we have the following result.

Theorem 2.3. Min-2-D-Partz is APX-hard.


Proposition 2.4 ([18]). If the unique games conjecture is true then Min-VC cannot
be approximated within (2 − ε), for any ε > 0.
The following lower bound result for Min-2-D-Partz follows from Theorem 2.3
and Proposition 2.4.

Theorem 2.5. If the unique games conjecture is true then Min-2-D-Partz cannot
be approximated within (2 − ε), for any ε > 0.
Now, we prove that Min-2-D-Partz has a two-factor approximation algorithm
and it is the best possible approximation factor because of Theorem 2.5. In polyno-
mial time it could be verified whether χd (G) ≤ 2 (Proposition 2.1). For a given graph
G, we want to compute a minimum size vertex set S such that χd (G[V \S]) ≤ 2.
If χd (G[V \S]) = 1 then a minimum size S can be computed in polynomial
time. It remains to compute a minimal 2-dominator partization set S such that
χd (G[V \S]) = 2, or G[V \S] is a biclique. In this case, Min-2-D-Partz is same
as minimum biclique deletion problem (Min-Biclique-Del). Given a graph G, in
Min-Biclique-Del, aim is to compute a vertex set S ⊆ V such that G[V \S] is a
biclique and the vertex set S is called biclique deletion set. Min-Biclique-Del is
a known NP-complete problem. There is a two-factor approximation algorithm for
Min-Biclique-Del based on an integer linear program formulation [16]. We design
a 2 factor approximation algorithm which is combinatorial in nature and is indepen-
dent of any integer linear program formulation. While designing this algorithm, we
consider a variation called minimum restricted biclique deletion problem (in short
Min-R-Biclique-Del) and prove similar hardness results. Min-R-Biclique-Del is
defined as, given a graph G = (V, E) and a vertex bipartition V = A ∪˙ B, it is

2250188-4
July 31, 2023 11:18 WSPC/S1793-8309 257-DMAA 2250188

Approximation algorithm for minimum q-dominator partization problem

required to find a subset S ⊆ V of minimum size such that G[V \S] is a biclique
with vertex bipartition X ∪˙ Y such that X ⊆ A and Y ⊆ B and the deletion set S
is called a restricted biclique deletion set.
Here, we prove that Min-Biclique-Del and Min-R-Biclique-Del are APX-
complete and they cannot be approximated within a factor smaller than 2. Our
approximation algorithm for Min-2-D-Partz is based on the approximation algo-
rithms for Min-Biclique-Del and Min-R-Biclique-Del.

Theorem 2.6. Min-R-Biclique-Del is APX-hard.

Proof. We prove this by an L-reduction from Min-VC to Min-R-Biclique-Del.


Given an instance G = (V, E) of Min-VC, in polynomial time, we construct a
graph G = (V  , E  ) and a vertex bipartition V  = A ∪˙ B as an instance of
Min-R-Biclique-Del as follows.
After making two copies G1 = (V1 , E1 ) and G2 = (V2 , E2 ) of G, we introduce n2
edges such that each vertex in V1 becomes adjacent to each vertex in V2 . Then, we
introduce a complete bipartite graph Kn,n with vertex sets X1 and X2 as partites.
Next we introduce another set of 2n2 edges such that each vertex in X1 becomes
adjacent to each vertex in V2 and each vertex in X2 becomes adjacent to each
vertex in V1 . Finally, we set a vertex bipartition of V  = A ∪˙ B with A = V1 ∪ X1
and B = V2 ∪ X2 .
If S is a minimal vertex cover in G then F = S1 ∪ S2 is a minimal biclique
deletion set in G , where S1 and S2 are corresponding copies of S in G1 and G2 ,
respectively. F is a minimal biclique deletion set in G because V1 \S1 and V2 \S2
are maximal independent sets in G1 and G2 , respectively, also G [V  \F ] is a maxi-
mal complete bipartite subgraph of G . Therefore, G [V  \F ] is biclique and F is a
biclique deletion set in G with |F | = 2|S|.
Before proving the converse part, we first claim that if F ∗ is a minimum biclique
deletion set in G then (X1 ∪ V1 )\F ∗ and (X2 ∪ V2 )\F ∗ are nonempty indepen-
dent sets in G and F ∗ ⊆ (V1 ∪ V2 ). Without loss of generality, let us assume that
(X1 ∪ V1 )\F ∗ is not an independent set in G . This implies that G1 [V1 \F ∗ ] contains
at least one edge (such a case can happen when G [V  \F ∗ ] is a complete bipartite
subgraph of G1 ). Then, (X2 ∪ V2 ) ⊆ F ∗ otherwise G [V  \F ∗ ] will contain a triangle.
Also, X1 ⊆ F ∗ as G [V  \F ∗ ] is a connected graph. This implies that |F ∗ | ≥ 3n.
This is a contradiction to the fact that size of a minimum biclique deletion set in
G can be at most 2n − 2, because (V1 \{x}) ∪ (V2 \{y}) is a biclique deletion set for
any x ∈ V1 and y ∈ V2 .
Based on these properties of a minimum biclique deletion set in G , we will
assume that if F is a minimal biclique deletion set in G then F ⊆ (V1 ∪ V2 ). Also,
V1 \F and V2 \F are independent sets in G1 and G2 , respectively. Let F = S1 ∪ S2
where S1 = V1 ∩ F and S2 = V2 ∩ F . Without loss of generality, we shall assume
that |S1 | = |S2 |. Suppose |S1 | < |S2 |, then it is easy to prove that F  = S1 ∪ S2 is a
minimal biclique deletion set in G , where S2 is the set of vertices in V2 corresponding

2250188-5
July 31, 2023 11:18 WSPC/S1793-8309 257-DMAA 2250188

S. Das & S. Mishra

to the vertices in S1 . Also |F  | < |F |. Now, from F we construct a vertex set S ⊆ V


as the set of vertices in V corresponding to the vertices in S1 . It is easy to observe
that |F | = 2|S| and S is a vertex cover in G.
Therefore, it is a L-reduction with α = 2 and β = 12 .

By using Proposition 2.4 and Theorem 2.6 we have the following lower bound
result.

Theorem 2.7. If the unique games conjecture is true then Min-R-Biclique-Del


cannot be approximated within (2 − ε), for any ε > 0.

Before describing our two-factor algorithm for Min-R-Biclique-Del, we first fix


some notation that we will use. Given a pair of disjoint vertex sets X and Y in G, we
define c(X, Y ) = {(u, v) ∈ E | u ∈ X and v ∈ Y } and c(X, Y ) = {(u, v) ∈ / E |u ∈ X
and v ∈ Y }. The input for Min-R-Biclique-Del is a graph G = (V, E) and a
vertex bipartition V = A ∪˙ B. For a cross edge (x, y) with x ∈ A and y ∈ B, we
define Nxy = N (x) ∩ N (y), N̂ (x) = (B ∩ N (x))\Nxy , N̂ (y) = (A ∩ N (y))\Nxy
and Txy = V \[N̂ (x) ∪ N̂ (y) ∪ Nxy ]. For each cross edge (x, y) ∈ E, our algorithm
computes a biclique subgraph Lxy = (VL , EL ) of G such that Lxy has vertex bipar-
tition VL = AL ∪˙ BL with AL ⊆ A, BL ⊆ B. The construction of biclique Lxy
associated with a cross edge (x, y) is as follows. We first construct a new graph
Hxy = (N̂ (x) ∪ N̂ (y), EH ) where EH = {(p, q) ∈ E | p, q ∈ N̂ (x)} ∪ {(p, q) ∈
E | p, q ∈ N̂ (y)} ∪ c(N̂ (x), N̂ (y)). Then, we compute a two-factor approximate ver-
tex cover Cxy in Hxy (using Proposition 2.2) and set Sxy = Nxy ∪ Txy ∪ Cxy as
a restricted biclique deletion set with respect to the edge (x, y). For the correct-
ness of the algorithm, it is enough to show that Lxy = G[V \Sxy ] is a biclique
satisfying the given vertex bipartition V = A ∪˙ B. We define AL = N̂ (y)\Cxy and
BL = N̂ (x)\Cxy . Since Cxy is a vertex cover in Hxy and EH contains the edge set
{(p, q) ∈ E | p, q ∈ N̂ (x)} ∪ {(p, q) ∈ E | p, q ∈ N̂ (y)}, AL and BL are independent
sets. At the same time, since Hxy contains the edge set c(N̂ (x), N̂ (y)) and Cxy is
a vertex cover, Cxy contains at least one end vertex of each edge of c(N̂ (x), N̂ (y)).
Therefore, Lxy is a biclique. We write this algorithm formally in Algorithm 1. By
Proposition 2.2, Cxy can be computed in polynomial time. All other computations
take constant time. Hence Algorithm 1 runs in polynomial time.

Theorem 2.8. Algorithm 1 approximates Min-R-Biclique-Del within a factor


of 2.

Proof. We denote the vertex deletion set associated with the cross edge (x, y) by

Mxy . We define Mxy as the minimum vertex deletion set associated with the cross

edge (x, y). Let F be a minimum size restricted biclique deletion set in the graph
G with respect to the vertex bipartition A ∪ B. Let (x∗ , y ∗ ) be a cross edge such
that F ∗ = Mx∗∗y∗ . Also, let Cx∗∗ y∗ be a minimum size vertex cover in Hx∗ y∗ . Let
Fapx be the solution of Min-R-Biclique-Del for G computed by Algorithm 1 and

2250188-6
July 31, 2023 11:18 WSPC/S1793-8309 257-DMAA 2250188

Approximation algorithm for minimum q-dominator partization problem

Algorithm 1: Approximation algorithm for min-R-Biclique-deletion.


Input: G = (V, E) with vertex bipartition V = A ∪˙ B
Output: Fapx ⊆ V such that G[V \Fapx ] is a biclique with vertex bipartition
(A\Fapx , B\Fapx )
Fapx = ∅; temp = n;
for every cross edge (x, y) ∈ E with x ∈ A and y ∈ B do
Nxy = N (x) ∩ N (y);
N̂ (x) = (B ∩ N (x))\Nxy , N̂ (y) = (A ∩ N (y))\Nxy ;
Txy = V \[N̂ (x) ∪ N̂ (y) ∪ Nxy ];
Hxy = (N̂ (x) ∪ N̂ (y), EH );
Compute a two-factor approximate vertex cover Cxy in Hxy ;
Mxy = Nxy ∪ Txy ∪ Cxy ;
if |Mxy | < temp then
Fapx = Mxy ;
temp = |Fapx |;
end
end
Return Fapx ;

(x , y  ) be the corresponding cross edge.


|Fapx | |Mx y | |Nx y | + |Tx y | + |Cx y |
= =
|F ∗ | |Mx∗∗ y∗ | |Nx∗ y∗ | + |Tx∗ y∗ | + |Cx∗∗ y∗ |
|Nx∗ y∗ | + |Tx∗y∗ | + |Cx∗ y∗ |

|Nx∗ y∗ | + |Tx∗y∗ | + |Cx∗∗ y∗ |
(∵ |Nx y | + |Tx y | + |Cx y | ≤ |Nx∗ y∗ | + |Tx∗ y∗ | + |Cx∗ y∗ |)
|Cx∗ y∗ |
≤ (∵ |Cx∗∗ y∗ | ≤ |Cx∗ y∗ |)
|Cx∗∗ y∗ |
≤ 2.

The last inequality follows from the fact that Cx∗ y∗ is a two-factor approximate
vertex cover in the graph Hx∗ y∗ .

Now we use Algorithm 1 to design algorithm for Min-Biclique-Del. Our algo-


rithm is based on the following idea. For each edge e = (x , y  ) in G, we compute the
sets Tx y = V \[N (x ) ∪ N (y  )], Nx y = N (x ) ∩ N (y  ), N̂ (x ) = N (x )\Nx y and
N̂ (y  ) = N (y  )\Nx y . Then, we consider the subgraph Ux y = G[N̂ (x ) ∪ N̂ (y  )] of
G. It is easy to observe that in Ux y , all the neighbors of x belongs to N̂ (x ) and
all the neighbors of y  belongs to N̂ (y  ). But Ux y need not be a bipartite graph.
By using Algorithm 1, we compute a restricted biclique deletion set Rx y in Ux y .

2250188-7
July 31, 2023 11:18 WSPC/S1793-8309 257-DMAA 2250188

S. Das & S. Mishra

Algorithm 2: Approximation algorithm for Min-Biclique-Deletion.


Input: G = (V, E)
Output: Sapx ⊆ V such that G[V \Sapx ] is a biclique
Sapx = ∅; temp = n;
for each edge (x , y  ) ∈ E do
Tx y = V \[N (x ) ∪ N (y  )];
Nx y = N (x ) ∩ N (y  );
N̂ (x ) = N (x )\Nx y , N̂ (y  ) = N (y  )\Nx y ;
Ux y = G[N̂ (x ) ∪ N̂ (y  )];
Compute a two-factor approximate restricted biclique deletion set Rx y
in Ux y with the vertex bipartition N̂ (x ) and N̂ (y  ) using Algorithm 1;
Ax y = Tx y ∪ Nx y ∪ Rx y ;
if |Ax y | < temp then
Sapx = Ax y ;
temp = |Sapx |;
end
end
Return Sapx ;

Finally, we compute Ax y = Tx y ∪ Nx y ∪ Rx y as the vertex deletion set associ-
ated with the edge (x , y  ). We write this algorithm formally in Algorithm 2. Rx y
can be computed in polynomial time by Algorithm 1. All other steps take constant
time to run. Hence Algorithm 2 runs in polynomial time.
Theorem 2.9. Min-Biclique-Del can be approximated within a factor of 2.

Proof. We denote the vertex deletion set associated with the edge (x , y  ) by Ax y .
We define A∗x y as the minimum vertex deletion set associated with the edge (x , y  ).
Let S ∗ be an optimal solution of Min-Biclique-Del for G. Let (x∗ , y ∗ ) be the edge
such that R∗ = A∗x∗ y∗ . Also, let Rx∗ ∗ y∗ be a minimum size restricted biclique deletion
set in Ux∗ y∗ . Let Sapx be the solution of Min-Biclique-Del for G computed by
Algorithm 2 and (x1 , y1 ) be the corresponding edge.
|Sapx | |Ax y | |Tx1 y1 | + |Nx1 y1 | + |Rx1 y1 |
= ∗1 1 =
|S ∗ | |Ax∗ y∗ | |Tx∗ y∗ | + |Nx∗ y∗ | + |Rx∗ ∗ y∗ |
|Tx∗ y∗ | + |Nx∗ y∗ | + |Rx∗ y∗ |

|Tx∗ y∗ | + |Nx∗ y∗ | + |Rx∗ ∗ y∗ |
(∵ |Tx1 y1 | + |Nx1 y1 | + |Rx1 y1 | ≤ |Tx∗ y∗ | + |Nx∗ y∗ | + |Rx∗ y∗ |)
|Rx∗ y∗ |
≤ (∵ |Rx∗ ∗ y∗ | ≤ |Rx∗ y∗ |)
|Rx∗ ∗ y∗ |
≤ 2.

2250188-8
July 31, 2023 11:18 WSPC/S1793-8309 257-DMAA 2250188

Approximation algorithm for minimum q-dominator partization problem

The last inequality follows from the fact that Rx∗ y∗ is a two-factor approximate
restricted biclique deletion set in the graph Ux∗ y∗ .

Theorem 2.10. Min-2-D-Partz can be approximated within a factor of 2.

Proof. Since we know that χd (G) = 2 if and only if G is a complete bipartite


graph (Proposition 2.1), so the vertex deletion set of Min-Biclique-Del is also a
vertex deletion set for Min-2-D-Partz. Hence, we can conclude that Algorithm 2
approximates Min-2-D-Partz within a factor of 2.

3. Minimum 3-Dominator-Partization
In this section, we show that Min-3-D-Partz is APX-hard and we design an

O( log n) factor approximation algorithm for it. A subset O ⊆ V is an odd cycle
transversal in G if O has a nonempty intersection with every odd cycle in G. In
Min-OCT it is required to find an odd cycle transversal of minimum size in a given
simple graph G.
Proposition 3.1 ([17]). If unique games conjecture is true then Min-OCT does not
admit a constant factor approximation algorithm.

√ Unless P = NP, Min-OCT cannot be approximated within a


Proposition 3.2 ([5]).
factor smaller than 8 5 − 15.

Proposition 3.3 ([1]). Min-OCT is approximated within a factor of O( log n).
The L-reduction from Min-OCT to minimum 3-domination partization given in
[9] is also an L-reduction from Min-OCT to Min-3-D-Partz. Hence, Min-3-D-Partz
is as hard as Min-OCT. Therefore, we have the following result.
Theorem 3.4. Min-3-D-Partz is APX-hard.
Since Min-3-D-Partz is as hard as Min-OCT, from Propositions 3.1 and 3.2, we
have the following lower bound results for Min-3-D-Partz.
Theorem 3.5. If Unique Games Conjecture is true then Min-3-D-Partz does not
admit a constant factor approximation algorithm.
√ Min-3-D-Partz is NP-hard to
approximate within a factor smaller than 8 5 − 15.

It is known that Min-OCT is approximable within a factor of O( log n) by

a polynomial time randomized algorithm [1]. We design an O( log n) factor
approximation algorithm for Min-3-D-Partz by using approximation algorithms
for Min-OCT and Min-VC. Our algorithm is based on a characterization of 3-
dominator colorable graphs [3] and approximation algorithms for Min-VC and
Min-OCT. Based on the characterizations of graphs with χd (G) = 2 (Proposition 2.1)
and graphs with χd (G) = 3 due to Chellali and Maffray [3], we have the following
proposition.
Proposition 3.6 ([3]). A connected graph G satisfies χd (G) ≤ 3 if and only if G
is one of the following types of graph classes:

2250188-9
July 31, 2023 11:18 WSPC/S1793-8309 257-DMAA 2250188

S. Das & S. Mishra

(i)F0 = {G = (V, E) | |V | ≤ 3, |E| = 0}.


(ii)F1 = {Kr,s | r ≥ 1, s ≥ 1}.
(iii)F2 = {Kr,s ∪ K1 | r ≥ 1, s ≥ 1}.
(iv) F3 is the collection of graphs G = (V, E) such that it has two non-adjacent ver-
tices x, y with V \{x, y} = N (x) ∪ N (y) is an independent set, N (x)\N (y) = ∅
and N (y)\N (x) = ∅ (N (x) ∩ N (y) may or may not be empty).
(v) F4 is the collection of graphs of the form H ⊕ S, where H is any bipartite
graph with at least one edge and S = K p , for any p ≥ 1.
(vi) A graph G is in the class F5 if its vertex set can be partitioned into five
independent sets X1 , X2 , X3 , X4 , Z such that
— X1 , X2 , X3 , X4 are nonempty sets,
— every vertex in Xi is adjacent to every vertex in Xi+1 , for i = 1, 2, 3,
— every vertex in Z is adjacent to every vertex in X2 ∪ X3 ,
— Z ∪ X1 ∪ X4 is an independent set,
— every vertex in X1 has a non-neighbor in X3 and every vertex in X2 has
a non-neighbor in X4 .
(vii) A graph G is in the class F6 if its vertex set can be partitioned into five
nonempty sets {x}, X1 , X2 , X3 , X4 such that
— every vertex in Xi is adjacent to every vertex in Xi+1 , for i = 1, 2, 3,
— X1 ∪ X3 and X2 ∪ X4 are independent sets,
— x is adjacent to every vertex in X1 ∪ X4 ,
— there may be edges between x and vertices in X2 ∪ X3 ,
— there may be edges between vertices in X1 and X4 .
(viii) A graph G is in the class F7 if its vertex set can be partitioned in to six
nonempty sets X1 , X2 , X3 , Y1 , Y2 , Y3 such that, for i = 1, 2, 3,
—Xi ∪ Yi is an independent set,
—every vertex in Xi is adjacent to all of Xi+1 ∪ Yi+1 (modulo 3),
—every vertex in Yi is adjacent to every vertex in Xi+2 ∪ Yi+2 (modulo 3),
—every vertex in Xi has a non-neighbor in Yi+2 (modulo 3).

For designing an O( log n) factor approximation algorithm for Min-3-D-Partz
problem, we need a two-factor approximation algorithm for a modified minimum
vertex cover problem called minimum restricted vertex cover problem (Min-R-VC).
Given a graph G and a pair of non-adjacent vertices u, v ∈ V with d(v) ≥ 2 and
d(u) ≥ 2, in Min-R-VC it is required to find a vertex cover S of minimum size such
that both u and v are not in S. We prove that Min-R-VC is approximable within a
factor 2.

Theorem 3.7. Min-R-VC is approximable within a factor of 2.

Proof. We prove this theorem by establishing an L-reduction from Min-R-VC to


Min-VC with α = 1 and β = 1.

2250188-10
July 31, 2023 11:18 WSPC/S1793-8309 257-DMAA 2250188

Approximation algorithm for minimum q-dominator partization problem

Given a graph G = (V, E) and a pair of non-adjacent vertices u, v ∈ V (an


instance of Min-R-VC) we construct a graph G = (V  , E  ) as follows. First delete the
vertex v from G. Then, for each pair of vertices vi , vj ∈ NG (v) (the set of neighbors
of v in G) introduce three vertices vij 1 2
, vij 3
, vij and six edges {(vi , vij t t
), (vij , vj ) | 1 ≤
t ≤ 3}. Similarly, we delete the vertex u and introduce the necessary set of vertices
and edges, for every pair of vertices ui , uj ∈ NG (u).
Now, we claim that G has a minimal vertex cover of size k that exclude u, v if
and only if G has a minimal vertex cover of size k.
Let S be a minimal vertex cover in G with u, v ∈ / S. Then NG (u) ∪ NG (v) ⊆ S
and hence S is a minimal vertex cover in G . Clearly, S contains at least one of the
end vertices of the edges in E  ∩ E. Also, for each edge in E  \E, S contains exactly
one vertex (this vertex is either in NG (u) or in NG (v)).
Conversely, let S be a minimal vertex cover in G . We assume that S does not
contain any vertex vij t
, where vi , vj ∈ NG (v) and t ∈ {1, 2, 3}. Otherwise, S must
1 2 3
contain all the three vertices vij , vij and vij and at least one of vi and vj is not
in S. Then, we construct a new vertex cover S  = S\{vij 1 2
, vij 3
, vij } ∪ {vi , vj } in G
with |S  | < |S|. Similarly, it can be proved that S does not contain any vertex utij ,
where ui , uj ∈ NG (u) and t ∈ {1, 2, 3}. From this observation it also follows that
NG (u) ∪ NG (v) ⊆ S. Hence, S is a subset of V and is a vertex cover in G.
Now it is easy to observe that this reduction is an L-reduction with α = 1 and
β = 1. This implies that Min-R-VC is approximable within a factor of 2.

Our objective is to compute a minimal 3-dominator partization set S of min-


imum size. If S is a minimal 3-dominator partization set in G then the graph
G[V \S] must be one of the eight types mentioned in Proposition 3.6. We compute
seven approximate deletion sets S0 , S2 , S3 , . . . , S7 corresponding to eight types of
3-dominator colorable graphs in polynomial time and then choose the deletion set
with minimum size among them. If G[V \S] is of Type 0 then computation of S0 is
trivial. In fact, an optimal deletion set S0 will satisfy the property that |S0 | = 0 if
|V | ≤ 3 and has no edge. If |V | ≥ 4 then |S0 | = |V | − |I| where I is an independent
set of size at most 3 in G.
For other types, we are designing separate algorithm for each type. In each
of these algorithms, we compute either approximate vertex covers or odd cycle
transversal in some related graphs. For a given set of vertices A in G, we define
E[A] as the set of edges of G with both end vertices in A.

3.1. Graph class F2


When G is a graph such that it has a minimum 3-dominator partization set S2∗
with the property that G[V \S2∗ ] ∈ F2 , then, in polynomial time we compute a
two-factor approximate 3-dominator partization set S2 such that G[V \S2 ] ∈ F2 .
Keeping in mind this structural property, we first take a vertex v ∈ V . Then we

2250188-11
July 31, 2023 11:18 WSPC/S1793-8309 257-DMAA 2250188

S. Das & S. Mishra

compute a minimal 2-dominator partization set Xv corresponding to v in the sub-


graph G[V \N [v]] (N [v] denotes the closed neighborhood of v), using Algorithm 2,
in polynomial time. Finally, we return Sv = Xv ∪ N (v) as a 3-dominator partization
set with respect to the vertex v. We do this step for each vertex v in G and take the
one with minimum size as our deletion set S2 . Formally we describe this algorithm in
Algorithm 3.

Algorithm 3: Deletion-type-2(G).
Input: G = (V, E) such that there exists a vertex set S with G[V \S] ∈ F2
Output: S2 ⊆ V such that G[V \S2 ] ∈ F2
S2 = ∅; temp = n;
for each vertex v ∈ V do
Construct the subgraph Gv = G[V \N [v]];
Compute a biclique deletion set Xv in G[V \N [v]] by using Algorithm 2;
Sv = Xv ∪ N (v);
if |Sv | < temp then
S2 = Sv ;
temp = |S2 |;
end
end
Return S2 ;

Lemma 3.8. Let G be a graph such that it has a minimum size 3-dominator par-
tization set S2∗ with G[V \S2∗ ] ∈ F2 . Then S2 = Deletion-type-2(G) is a 2 factor
approximate solution.

Proof. Let S2∗ be a minimum size 3-dominator partization set in G such that
G[V \S2∗ ] ∈ F2 . Let v ∗ be the vertex that forms the component with only one
vertex and let Xv∗∗ be the minimum biclique deletion set in G[V \N [v ∗ ]]. Let Xv∗∗ =
Tx∗y∗ ∪ Nx∗ y∗ ∪ Rx∗ ∗ y∗ as described in the proof of Theorem 2.9. By following the
notations from the proof of Theorem 2.9, we can write S2 = N (v) ∪ Txy ∪ Nxy ∪ Sxy
for some edge (x, y) in G[V \N [v]].
|S2 | |N (v)| + |Txy | + |Nxy | + |Rxy |
=
|S2∗ | |N (v ∗ )| + |Tx∗ y∗ | + |Nx∗ y∗ | + |Rx∗ ∗ y∗ |
|N (v ∗ )| + |Tx∗ y∗ | + |Nx∗ y∗ | + |Rx∗ y∗ |

|N (v ∗ )| + |Tx∗ y∗ | + |Nx∗ y∗ | + |Rx∗ ∗ y∗ |
(∵ |N (v)| + |Txy | + |Nxy | + |Rxy | ≤ |N (v ∗ )|
+ |Tx∗y∗ | + |Nx∗ y∗ | + |Rx∗ y∗ |)

2250188-12
July 31, 2023 11:18 WSPC/S1793-8309 257-DMAA 2250188

Approximation algorithm for minimum q-dominator partization problem

|Rx∗ y∗ |
≤ (∵ |Rx∗ ∗ y∗ | ≤ |Rx∗ y∗ |)
|Rx∗ ∗ y∗ |
≤ 2.
The last inequality follows from the fact that Rx∗ y∗ is a 2 factor approximate
restricted biclique deletion set in the graph Ux∗ y∗ .

3.2. Graph class F3


Based on the structural properties of the graph class F3 , for every pair of non-
adjacent vertices x, y in G, we compute the vertex set Txy = V \(N [x] ∪ N [y]) and
a restricted vertex cover Sxy in the subgraph G[N (x) ∪ N (y)]. For every pair of
vertices p ∈ N (x)\N (y) and q ∈ N (y)\N (x), we compute a two-factor approximate
restricted vertex cover Spq , by using Theorem 3.7 in polynomial time. Finally, we
choose a set of minimum size among these restricted vertex covers. We shall denote
this set as Sxy . Now Pxy = Txy ∪ Sxy is a 3-dominator partization set in G with
respect to the vertex pair x and y. In the last step, the algorithm selects a 3-
dominator partization set S3 such that |S3 | = min{|Pxy | | x and y are non-adjacent
pair of vertices in V }. Computing Txy , Pxy , S3 and selecting Sxy among all Spq
takes constant time. We will denote this algorithm as Deletion-type-3(G) and its
formal description is available in Algorithm 4 which runs in polynomial time.

Lemma 3.9. Let G be a graph such that it has a minimum size 3-dominator par-
tization set S3∗ with G[V \S3∗ ] ∈ F3 . Then S3 = Deletion-Type-3(G) is a two-factor
approximate solution.

Proof. Let S3∗ = Tx∗ y∗ ∪ Sx∗∗ y∗ be a minimum 3-dominator partization set and the
corresponding non-adjacent pair of vertices be x∗ and y ∗ . Let p∗ ∈ N (x∗ )\N (y ∗ )
and q ∗ ∈ N (y ∗ )\N (x∗ ) such that Sx∗∗ y∗ = Sp∗∗ q∗ and Sp∗∗ q∗ is a minimum restricted
vertex cover in the subgraph G[N (x∗ ) ∪ N (y ∗ )]. Now
|S3 | |Txy | + |Sxy | |Txy | + |Sxy | |Tx∗ y∗ | + |Sp∗ q∗ |
= = ≤
|S3∗ | |Tx∗ y∗ | + |Sx∗∗ y∗ | |Tx∗ y∗ | + |Sp∗∗ q∗ | |Tx∗ y∗ | + |Sp∗∗ q∗ |
(∵ |Txy | + |Sxy | ≤ |Tx∗ y∗ | + |Sx∗ y∗ | ≤ |Tx∗ y∗ | + |Sp∗ q∗ |)
|Sp∗ q∗ |
≤ (∵ |Sp∗∗ q∗ | ≤ |Sp∗ q∗ |)
|Sp∗∗ q∗ |
≤2 (∵ Sp∗ q∗ is a two-factor approximate solution for Min-R-VC).

3.3. Graph class F1 ∪ F4


Due to some technical issues, we combine both Types 1 and 4 together for designing
algorithms for these two graph classes. Let us assume G be a graph such that it has

2250188-13
July 31, 2023 11:18 WSPC/S1793-8309 257-DMAA 2250188

S. Das & S. Mishra

Algorithm 4: Deletion-type-3(G).
Input: G = (V, E) such that there exists a vertex set S with G[V \S] ∈ F3
Output: S3 ⊆ V such that G[V \S3 ] ∈ F3
S3 = ∅; temp = n;
for each pair of non-adjacent vertices x, y ∈ V do
Txy = V \[N [x] ∪ N [y]];
for each pair of vertices p ∈ N (x)\N (y) and q ∈ N (y)\N (x), compute a
two-factor approximate restricted vertex cover Spq in G[N (x) ∪ N (y)]
with respect to vertices p and q;
Select a restricted vertex cover Sxy such that it is of least size among the
restricted vertex covers Spq in G[N (x) ∪ N (y)];
Pxy = Txy ∪ Sxy ;
if |Pxy | < temp then
S3 = Pxy ;
temp = |S3 |;
end
end
Return S3 ;

a 3-dominator partization set S with G[V \S] ∈ F1 ∪ F4 . Based on the structural


properties of the graph class F1 and F4 , for each ordered pair < x, y > of adjacent
vertices x, y ∈ V , we compute a 3-dominator partization set Dxy as follows. First,
we define a gadget Juv = (VJuv , EJuv ) associated with an edge (u, v) ∈ E, where
VJuv = {u, v, uv1 , uv2 , uv3 } and
EJuv = {(u, v), (u, uv1 ), (u, uv2 ), (u, uv3 ), (uv1 , v), (uv2 , v), (uv3 , v)}.
For such an ordered pair x, y , construct the sets Txy = V \(N [x] ∪ N [y]), B =
N (x) and A = N (y)\N (x) in constant time. Next, we construct the graph Hxy =
(Vxy , Exy ) with
⎡ ⎤

Vxy = B ∪ ⎣ VJuv ⎦ and
(u,v)∈[E[A] ∪ c(A,B)]
⎡ ⎤

Exy = E[B] ∪ ⎣ EJuv ⎦.
(u,v)∈[E[A] ∪ c(A,B)]

Then we find an O( log n) factor odd cycle transversal Oxy in Hxy by Proposi-
tion 3.3 in polynomial time. From the structure of Juv , it can be observed that
at least one of u and v must be in any minimal odd cycle transversal. Therefore,
we assume that Oxy ⊆ (A ∪ B). Since, for each edge (u, v) in E[A] ∪ c(A, B), Hxy
contains three odd cycles with a common edge (u, v) and Oxy contains either u, v

2250188-14
July 31, 2023 11:18 WSPC/S1793-8309 257-DMAA 2250188

Approximation algorithm for minimum q-dominator partization problem

or both, A\Oxy is an independent set and each vertex in A\Oxy is adjacent to each
vertex in B\Oxy in G. Observe that, G[B\Oxy ] is a bipartite graph. If G[B\Oxy ]
contains at least one edge then G[V \Dxy ] ∈ F4 , otherwise G[V \Dxy ] ∈ F1 . Hence,
Dxy = Txy ∪ Oxy is 3-dominator partization set in G with respect to the ordered
pair x, y . Finally we select the one with minimum size as among all Dxy as S4 .
For a formal description, we refer to Algorithm 5. Since computing Oxy takes poly-
nomial time and other steps take constant time, Algorithm 5 runs in polynomial
time.

Algorithm 5: Deletion-type-1&4(G).
Input: G = (V, E) such that there exists a vertex set S with
G[V \S] ∈ F1 ∪ F4
Output: S4 ⊆ V such that G[V \S4 ] ∈ F1 ∪ F4
S4 = ∅; temp = n;
for each ordered pair < x, y > of adjacent vertices x, y ∈ V do
Txy = V \[N [x] ∪ N [y]];

Construct the graph Hxy and compute a O( log n) factor odd cycle
transversal Oxy in it;
Sxy = Txy ∪ Oxy ;
if |Sxy | < temp then
S4 = Sxy ;
temp = |S4 |;
end
end
Return S4 ;

Lemma 3.10. Let G be a graph such that it has a minimum size 3-dominator
partization set S4∗ with G[V \S4∗ ] ∈ F1 ∪ F4 . Then S4 = Deletion-Type-1&4(G) is a

O( log n) factor approximate solution.

Proof. Let S4∗ = Tx∗ y∗ ∪ Ox∗∗ y∗ be a minimum 3-dominator partization set and
the corresponding order pair < x∗ , y ∗ > of adjacent pair of vertices be x∗ and y ∗ .

Here, Ox∗∗ y∗ and Ox∗ y∗ are a minimum size odd cycle transversal and a O( log n)-
approximate odd cycle transversal in the subgraph G[N [x∗ ] ∪ N [y ∗ ]], respectively.
Now
|S4 | |Txy | + |Oxy |
=
|S4∗ | |Tx∗ y∗ | + |Ox∗∗ y∗ |
|Tx∗ y∗ | + |Ox∗ y∗ |

|Tx∗ y∗ | + |Ox∗∗ y∗ |
(∵ |Txy | + |Oxy | ≤ |Tx∗ y∗ | + |Ox∗ y∗ |)

2250188-15
July 31, 2023 11:18 WSPC/S1793-8309 257-DMAA 2250188

S. Das & S. Mishra

|Ox∗ y∗ |
≤ (∵ |Ox∗∗ y∗ | ≤ |Ox∗ y∗ |)
|Ox∗∗ y∗ |
 
≤ O( log n) (∵ Ox∗ y∗ is a O( log n) factor approximate solution).

3.4. Graph class F5


From the definition of F5 it follows that G has a 3-dominator partization set S
with G[V \S] ∈ F5 if and only if G has an induced P4 . We compute an approximate
3-dominator partization set SP4 in G with respect to each induced P4 in G, and
finally return the one with minimum size. Here we use minimum weight vertex
cover problem (Min-W-VC) and it’s known two-factor approximation algorithm, to
compute our deletion set. Given a graph G = (V, E) and a vertex weight function
w : V → N+ , in Min-W-VC it is required to find a vertex cover S of minimum weight

w(S) = v∈S w(v).

Proposition 3.11 ([22]). Min-W-VC can be approximated within a factor of 2.

Let P4 = x1 , x2 , x3 , x4 be an induced path in G. It is important to note that


the edges (x1 , x3 ) and (x2 , x4 ) are not in G. We construct the vertex sets as follows.

X̂2 = N (x1 ) ∩ N (x3 ),


X̂3 = N (x2 ) ∩ N (x4 ),
X̂1 = N (x2 )\(X̂2 ∪ X̂3 ),
X̂4 = N (x3 )\(X̂2 ∪ X̂3 ),
TP4 = V \[X̂1 ∪ X̂2 ∪ X̂3 ∪ X̂4 ].

Based on these sets, we construct a vertex weighted graph HP4 =


(VH , EH ) with VH = X̂1 ∪ X̂2 ∪ X̂3 ∪ X̂4 and EH = E[X̂1 ] ∪ E[X̂2 ] ∪ E[X̂3 ] ∪
E[X̂4 ] ∪ c(X̂1 , X̂4 ) ∪ c(X̂1 , X̂2 ) ∪ c(X̂2 , X̂3 ) ∪ c(X̂3 , X̂4 ). We set the weights of the
vertices as w(x1 ) = w(x2 ) = w(x3 ) = w(x4 ) = n and w(v) = 1 for all
v ∈ V \{x1 , x2 , x3 , x4 }. Then, we compute a two-factor approximate weighted vertex
cover CP4 in HP4 by Proposition 3.11 and set SP4 = TP4 ∪ CP4 as a 3-dominator
partization set associated with the P4 . We formally describe this algorithm in Algo-
rithm 6.
Next, we claim that SP4 is a 3-dominator partization set in G with G[V \SP4 ] ∈
F5 . We define X1 = X̂1 \CP4 , X2 = X̂2 \CP4 , X3 = X̂3 \CP4 and X4 = X̂4 \CP4 .
Since the vertices x1 , . . . , x4 have large weights, the vertex cover CP4 does not con-
tain these vertices. Therefore, x1 ∈ X1 , x2 ∈ X2 , x3 ∈ X3 and x4 ∈ X4 . Since
4
HP4 contains the edges in i=1 E[X̂i ] ∪ c(X̂1 , X̂4 ) and CP4 is a vertex cover it fol-
 
lows that X1 ∪ X4 , X2 and X3 are independent sets in G. Also, the subgraphs
G[X1 ∪ X2 ], G[X2 ∪ X3 ], G[X3 ∪ X4 ] are bicliques as HP4 contains the edges in
c(X̂1 , X̂2 ), c(X̂2 , X̂3 ), c(X̂3 , X̂4 ) and CP4 is a vertex cover in HP4 . Now, we define the

2250188-16
July 31, 2023 11:18 WSPC/S1793-8309 257-DMAA 2250188

Approximation algorithm for minimum q-dominator partization problem

Algorithm 6: Deletion-type-5(G).
Input: G = (V, E) such that there exists a vertex set S with G[V \S] ∈ F5
Output: S5 ⊆ V such that G[V \S5 ] ∈ F5
S5 = ∅; temp = n;
for each induced path P4 =< x1 , x2 , x3 , x4 > do
Construct the vertex weighted graph HP4 associated with P4 ;
Compute 2 factor weighted vertex cover CP4 in the graph HP4 ;
Construct the set SP4 = TP4 ∪ CP4 ;
if |SP4 | < temp then
S5 = SP4 ;
temp = |S5 |;
end
end
Return S5 ;

set Z = {v | v ∈ (X1 ∪ X4 ) and v is adjacent to every vertex in (X2 ∪ X3 )}. Finally,
we define the sets X1 = X1 \Z and X4 = X4 \Z. Now, the sets X1 , X2 , X3 , X4 and
Z satisfies the conditions for the graph class F5 .
Construction of the weighted graph HP4 and computation of the weighted vertex
cover CP4 take polynomial time. Other steps take constant time. Hence Algorithm 6
runs in polynomial time.
Lemma 3.12. Let G be a graph such that it has a minimum size 3-dominator
partization set S5∗ with G[V \S5∗ ] ∈ F5 . Then S5 = Deletion-type-5(G) is a two-
factor approximate solution.

Proof. We have mentioned that G has a 3-dominator partization set S with


G[V \S] ∈ F5 if and only if G has an induced P4 . Let S5∗ be a minimum size 3-
dominator partization set in G and let P4∗ be an induced path with four vertices
associated with S5∗ . Then S5∗ = TP4∗ ∪ CP∗4∗ with CP∗4∗ is a minimum weight vertex
cover in the vertex weighted graph HP4∗ . Let SP4 be set computed by Algorithm 6
and P4 be the associated induced path on four vertices. Now
|S5 | |TP4 | + |CP4 | |TP4∗ | + |CP4∗ |
= ≤
|S5∗ | |TP4∗ | + |CP∗ ∗ | |TP4∗ | + |CP∗ ∗ |
4 4

(∵ |TP4 | + |CP4 | ≤ |T P4∗ | + |CP4∗ |)


|CP4∗ |
≤ (∵ |CP∗4∗ | ≤ |CP4∗ |)
|CP∗ ∗ |
4

≤ 2.
The last inequality follows from the fact that CP4∗ is a two-factor approximate
weighted vertex cover in the graph HP4∗ .

2250188-17
July 31, 2023 11:18 WSPC/S1793-8309 257-DMAA 2250188

S. Das & S. Mishra

3.5. Graph class F6


Let us assume that G is a graph such that it has a minimum 3-dominator partization
set S6∗ with the property that G[V \S6∗ ] ∈ F6 . Let P M5 =< x, x1 , x2 , x3 , x4 > is
an ordered set of five distinct vertices such that G[{x, x1 , x2 , x3 , x4 }] contains the
cycle C5 = x, x1 , x2 , x3 , x4 , x and does not contain the pair of edges (x1 , x3 ) and
(x2 , x4 ) (pentagon without a particular matching). From the description of F6 it
follows that G has a 3-dominator partization set S with G[V \S] ∈ F6 if and only
if G has P M5 . We compute an approximate 3-dominator partization set DP M5 in
G with respect to each P M5 in G, and finally return the one with minimum size as
S6 . Here also we use Min-W-VC to compute our deletion set.
For a P M5 , first we construct the following five sets.
X̂2 = N (x1 ) ∩ N (x3 ),
X̂3 = N (x2 ) ∩ N (x4 ),
X̂1 = [N (x2 ) ∩ N (x)]\[X̂2 ∪ X̂3 ],
X̂4 = [N (x3 ) ∩ N (x)]\[X̂2 ∪ X̂3 ],
TP M5 = V \[X̂1 ∪ X̂2 ∪ X̂3 ∪ X̂4 ∪ {x}].
Based on these sets, we construct a vertex weighted graph HP M5 =
4 4
(VP M5 , EP M5 ) with VP M5 = i=1 X̂i ∪ {x} and EP M5 = i=1 E[X̂i ] ∪ c(X̂1 , X̂3 ) ∪
c(X̂2 , X̂4 ) ∪ c(X̂1 , X̂2 ) ∪ c(X̂2 , X̂3 ) ∪ c(X̂3 , X̂4 ). The vertex weight function is defined
as w(v) = n, if v ∈ {x, x1 , x2 , x3 , x4 } and w(v) = 1, if v ∈ [VP M5 \{x, x1 , x2 , x3 , x4 }].
Let CP M5 be a two-factor approximate weighted vertex cover in the vertex weighted
graph HP M5 . Finally, DP M5 = TP M5 ∪ CP M5 is a 3-dominator partization set with
G[V \DP M5 ] ∈ F6 . The formal description is given in Algorithm 7.
Now, we claim that DP M5 is a 3-dominator partization set in G with
G[V \DP M5 ] ∈ F6 . We define the sets Xi = X̂i \CP M5 , for i = 1, . . . , 4. Since
the vertices x, x1 , . . . , x4 are assigned with a larger weight these vertices are not
in the vertex cover CP M5 . As a result of which xi ∈ Xi , for i = 1, . . . , 4. Since
CP M5 contains at least one end vertices of the edges in 4i=1 E[X̂i ], each Xi is an
independent set. Inclusion of the edge set c(X̂1 , X̂3 ) ∪ c(X̂2 , X̂4 ) in HP H5 implies
that there is no cross edge between the vertex sets X1 and X3 , and between X2 and
X4 . As HP H5 contains the edge set c(X̂1 , X̂2 ) ∪ c(X̂2 , X̂3 ) ∪ c(X̂3 , X̂4 ), the induced
subgraphs G[X1 ∪ X2 ], G[X2 ∪ X3 ] and G[X3 ∪ X4 ] are bicliques.

Lemma 3.13. Let G be a graph such that it has a minimum size 3-dominator
partization set S6∗ with G[V \S6∗ ] ∈ F6 . Then S6 = Deletion-type-6(G) is a 2 factor
approximate solution.

Proof. We have mentioned that G has a 3-dominator partization set S with


G[V \S] ∈ F6 if and only if G has an induced P M5 . Let S5∗ be a minimum size
3-dominator partization set in G and let P M5∗ be an ordered set of five vertices

2250188-18
July 31, 2023 11:18 WSPC/S1793-8309 257-DMAA 2250188

Approximation algorithm for minimum q-dominator partization problem

Algorithm 7: Deletion-type-6(G).
Input: G = (V, E) such that there exists a vertex set S with G[V \S] ∈ F6
Output: S6 ⊆ V such that G[V \S6 ] ∈ F6
S6 = ∅; temp = n;
for each P M5 = x, x1 , x2 , x3 , x4 do
Construct the vertex weighted graph HP M5 associated with P M5 ;
Compute two-factor vertex cover CP M5 in the graph HP M5 ;
Construct the set DP M5 = TP M5 ∪ CP M5 ;
if |DP M5 | < temp then
S6 = DP M5 ;
temp = |S6 |;
end
end
Return S6 ;

associated with S6∗ . Then S6∗ = TP M5∗ ∪ CP∗ M5∗ with CP∗ M5∗ is a minimum weight
vertex cover in the vertex weighted graph HP M5∗ . Let DP M5 be set computed by
Algorithm 7 and P M5 be the associated ordered set. Now

|S6 | |TP M5 | + |CP M5 | |TP M5∗ | + |CP M5∗ |


∗ = ∗ ≤
|S6 | |TP M5 | + |CP M ∗ |
∗ |TP M5∗ | + |CP∗ M ∗ |
5 5

(∵ |TP M5 | + |CP M5 | ≤ |T P M5∗ | + |CP M5∗ |)


|CP M5∗ |
≤ (∵ |CP∗ M5∗ | ≤ |CP M5∗ |)
|CP∗ M ∗ |
5

≤ 2.

The last inequality follows from the fact that CP M5∗ is a two-factor approximate
weighted vertex cover in the graph HP M5∗ .

3.6. Graph class F7


Now let us consider the graph G such that it has a minimum 3-dominator
partization set S7∗ with the property that G[V \S7∗ ] ∈ F7 . From the definition
of the graph class F7 it follows that G ∈ F7 if and only if G contains an
ordered set J6 = x1 , x2 , x3 , y1 , y2 , y3 of six vertices such that G[J6 ] has exactly
nine edges and this edge set is {(x1 , x2 ), (x2 , x3 ), (x1 , x3 ), (y1 , y2 ), (y2 , y3 ), (y1 , y3 ),
(x1 , y2 ), (x2 , y3 ), (x3 , y1 )}. We compute an approximate 3-dominator partization set
SJ6 in G with respect to each J6 in G, and finally return the one with minimum
size.

2250188-19
July 31, 2023 11:18 WSPC/S1793-8309 257-DMAA 2250188

S. Das & S. Mishra

For a J6 we compute the following sets:


Ŷ1 = [N (y2 ) ∩ N (y3 ) ∩ N (x3 )]\N (x1 ),
Ŷ2 = [N (y1 ) ∩ N (y3 ) ∩ N (x1 )]\N (x2 ),
Ŷ3 = [N (y1 ) ∩ N (y2 ) ∩ N (x2 )]\N (x3 ),
X̂1 = [N (x2 ) ∩ N (x3 ) ∩ N (y2 )]\N (y1 ),
X̂2 = [N (x1 ) ∩ N (x3 ) ∩ N (y3 )]\N (y2 ),
X̂3 = [N (x1 ) ∩ N (x2 ) ∩ N (y1 )]\N (y3 ),
TJ6 = V \[X̂1 ∪ X̂2 ∪ X̂3 ∪ Ŷ1 ∪ Ŷ2 ∪ Ŷ3 ].
Based on these vertex sets, we construct a vertex weighted graph HJ6 = (VH , EH ),
with VH = X̂1 ∪ X̂2 ∪ X̂3 ∪ Ŷ1 ∪ Ŷ2 ∪ Ŷ3 and EH = E1 ∪ E2 ∪ E3 , where

3 
3
E1 = E[X̂i ] ∪ E[Ŷi ] ,
i=1 i=1

E2 = c(X̂1 , Ŷ1 ) ∪ c(X̂2 , Ŷ2 ) ∪ c(X̂3 , Ŷ3 ) and


E3 = c(X̂1 , X̂2 ) ∪ c(X̂2 , X̂3 ) ∪ c(X̂3 , X̂1 ) ∪ c(Ŷ1 , Ŷ2 ) ∪ c(Ŷ2 , Ŷ3 )
∪ c(Ŷ3 , Ŷ1 ) ∪ c(X̂1 , Ŷ2 ) ∪ c(X̂2 , Ŷ3 ) ∪ c(X̂3 , Ŷ1 ).
The vertex weight function is defined as w(v) = n, if v ∈ J6 , and w(v) = 1, if
v ∈ VH \J6 . Let CJ6 be a two-factor vertex cover in HJ6 . We set SJ6 = TJ6 ∪ CJ6 as
the 3-dominator partization set in G with respect to J6 . We express this algorithm
formally in Algorithm 8.

Algorithm 8: Deletion-type-7(G).
Input: G = (V, E) such that there exists a vertex set S with G[V \S] ∈ F7
Output: S7 ⊆ V such that G[V \S7 ] ∈ F7
S7 = ∅; temp = n;
for each ordered set J6 =< a1 , a2 , a3 , b1 , b2 , b3 > do
Construct the vertex weighted graph HJ6 associated with J6 ;
Compute two-factor vertex cover CJ6 in the graph HJ6 ;
Construct the set SJ6 = TJ6 ∪ CJ6 ;
if |SJ6 | < temp then
S7 = SJ 6 ;
temp = |S7 |;
end
end
Return S7 ;

2250188-20
July 31, 2023 11:18 WSPC/S1793-8309 257-DMAA 2250188

Approximation algorithm for minimum q-dominator partization problem

Next, we claim that G[V \SJ6 ] ∈ F7 . We define the sets Xi = X̂i \CJ6 , for
i = 1, 2, 3 and Yi = Ŷi \CJ6 , for i = 1, 2, 3. Since the vertices in J6 are assigned with
a larger weight these vertices are not in the vertex cover CJ6 . As a result of which
xi ∈ Xi and yi ∈ Yi , for i = 1, 2, 3. Since CJ6 contains at least one end vertices of
the edges in E1 , the sets X1 , X2 , X3 , Y1 , Y2 , Y3 are independent sets. Inclusion of the
edge set E2 in HJ6 implies that the sets X1 ∪ Y1 , X2 ∪ Y2 and X3 ∪ Y3 are indepen-
dent sets. At the same time, inclusion of the edge set E3 implies that the induced
subgraphs G[X1 ∪ X2 ], G[X2 ∪ X3 ], G[X1 ∪ X3 ], G[Y1 ∪ Y2 ], G[Y2 ∪ Y3 ], G[Y1 ∪ Y3 ],
G[X1 ∪ Y2 ], G[X2 ∪ Y3 ] and G[X3 ∪ Y1 ] are bicliques. Now, if there exists a vertex
x1 ∈ X1 such that it is adjacent to all the vertices in Y3 , then we exclude x1 from
X1 and include it in Y1 . After shifting all such vertices from X1 to Y1 , we will
denote the resulting vertex sets as X1 and Y1 . Similarly, we construct the vertex
sets X2 , Y2 , X3 , Y3 . Now, it is easy to verify that G[{ 3i=1 Xi } ∪ { 3i=1 Yi }] ∈ F6 .

Lemma 3.14. Let G be a graph such that it has a minimum size 3-dominator
partization set S7∗ with G[V \S7∗ ] ∈ F7 . Then S7 = Deletion-type-7(G) is a two-
factor approximate solution.

Proof. We have mentioned that G has a 3-dominator partization set S with


G[V \S] ∈ F7 if and only if G has an induced J6 . Let S7∗ be a minimum size
3-dominator partization set in G and let J6∗ be an ordered set of six vertices asso-
ciated with S7∗ . Then S7∗ = TJ6∗ ∪ CJ∗6∗ with CJ∗6∗ is a minimum weight vertex cover
in the vertex weighted graph HJ6∗ . Let SJ6 be set computed by Algorithm 8 and J6
be the associated ordered set of six vertices. Now
|S7 | |TJ6 | + |CJ6 | |TJ6∗ | + |CJ6∗ |
∗ = ∗ ≤
|S7 | |TJ6∗ | + |CJ ∗ | |TJ6∗ | + |CJ∗∗ |
6 6

(∵ |TJ6 | + |CJ6 | ≤ |TJ6∗ | + |CJ6∗ |)


|CJ6∗ |
≤ (∵ |CJ∗6∗ | ≤ |CJ6∗ |)
|CJ∗∗ |
6

≤ 2.

The last inequality follows from the fact that CJ6∗ is a two-factor approximate
weighted vertex cover in the graph HJ6∗ .

3.7. Algorithm for Min-3-D-Partz


Now we design a polynomial time approximation algorithm for Min-3-D-Partz.
We use the polynomial time algorithms Algorithms 3–8 to design our algorithm.
We describe our algorithm in Algorithm 9.

Theorem 3.15. Algorithm 9 approximates Min-3-D-Partz within a factor of



O( log n).

2250188-21
July 31, 2023 11:18 WSPC/S1793-8309 257-DMAA 2250188

S. Das & S. Mishra

Algorithm 9: Algorithm for Min-3-D-Partz.


Input: A graph G = (V, E)
Output: S ⊆ V such that G[V \S] is 3-dominator-colorable.
if |V | ≤ 3 then
S0 = ∅;
end
else
S0 = V \{x, y, z} such that G[{x, y, z}] has no edge;
/* where x, y, z are any three vertices*/
end
Compute S2 using Algorithm 3;
Compute S3 using Algorithm 4;
Compute S4 using Algorithm 5;
Compute S5 using Algorithm 6;
Compute S6 using Algorithm 7;
Compute S7 using Algorithm 8;
S = argmin{|S0 |, |S2 |, |S3 |, |S4 |, |S5 |, |S6 |, |S7 |};
Return S;

Proof. Let S0 , S2 , S3 , . . . , S7 be the approximate deletion sets computed by the


algorithms described in this section. Let S2∗ , S3∗ , . . . , S7∗ be the minimum size 3-
dominator partization sets in respective graph classes.
|Si | |Sj |
∗ ≤ ∗ (as |Si | ≤ |Sj |)
|Sj | |Sj |

≤ O( log n).

The last inequality is because of Lemmas 3.8–3.10, 3.12–3.14.

4. Conclusion
We proved that Min-2-D-Partz is approximable with a factor of two
and it is the best possible approximation factor. Given a graph G, in
Max-2-Dominator-Partization it is required to find a vertex set S of maxi-
mum size such that G[S] is a 2-dominator-colorable graph. It can be shown that
Max-2-Dominator-Partization is equivalent to Maximum Independent Set and
cannot be approximable within a factor better than O(n1−ε ), for any ε > 0, unless
P = NP, because Maximum Independent Set has such a lower bound [15]. We also
prove that Min-3-D-Partz is equivalent to Min-OCT and is approximable within a

factor of O( log n). We expect that Min-3-D-Partz cannot be approximable within
a factor better than O(n1−ε ), for any ε > 0, unless P = NP.

2250188-22
July 31, 2023 11:18 WSPC/S1793-8309 257-DMAA 2250188

Approximation algorithm for minimum q-dominator partization problem

References

[1] A. Agarwal, M. Charikar, K. Makaryche and Y. Makarychev, O( log n) approxi-
mation algorithms for min uncut, min 2cnf deletion, and directed cut problems, in
Proc. 37th Annual ACM Symp. Theory of Computing (Baltimore, MD, USA, 2005),
pp. 573–581.
[2] S. Arumugam, K. R. Chandrasekar, N. Misra, G. Philip and S. Saurabh, Algorith-
mic aspects of dominator colorings in graphs, in Int. Workshop on Combinatorial
Algorithms (Springer, 2011), pp. 19–30.
[3] M. Chellali and F. Maffray, Dominator colorings in some classes of graphs, Graphs
Combin. 28(1) (2012) 97–107.
[4] Y. H. Chen, The dominated coloring problem and its application, in Int. Conf. Com-
putational Science and its Applications (Springer, 2014), pp. 132–145.
[5] M. Chlebı́k and J. Chlebı́ková, On approximation hardness of the minimum 2sat-
deletion problem, in Int. Symp. Mathematical Foundations of Computer Science
(Springer, 2004), pp. 263–273.
[6] V. Chvatal, A greedy heuristic for the set-covering problem, Math. Operations Res.
4(3) (1979) 233–235.
[7] S. Das, Complexities of vertex coloring problems with some domination properties,
PhD thesis, Indian Institute of Technology Madras (2021).
[8] S. Das and S. Mishra, Lower bounds on approximating some variations of vertex
coloring problem over restricted graph classes, Discrete Math. Algorithms Appl. 12(6)
(2020) 2050086.
[9] S. Das and S. Mishra, On the complexity of minimum q-domination partization prob-
lems, J. Combin. Opt. 43(2) (2022) 363–383.
[10] I. Dinur and D. Steurer, Analytical approach to parallel repetition, in Proc. 46th
Annual ACM Symp. Theory of Computing (New York, United States, 2014), pp. 624–
633.
[11] F. Gavril and M. Yannakakis, Independent unpublished work mentioned in Com-
binatorial Optimization: Algorithms and Complexity, C. H. Papadimitriou and K.
Steiglitz, (Courier Corporation, 2001), p. 432.
[12] R. Gera, On dominator colorings in graphs, Graph Theory Notes New York 52 (2007)
25–30.
[13] R. Gera, On the dominator colorings in bipartite graphs, in 4th Int. Conf. Information
Technology (IEEE, 2007), pp. 947–952.
[14] R. Gera, C. W. Rasmussen and S. Horton, Dominator colorings and safe clique par-
titions, Congress 181 (2006) 19–32.
[15] J. Håstad, Clique is hard to approximate within n1−ε , Acta Math. 182(1) (1999)
627–636.
[16] D. S. Hochbaum, Approximating clique and biclique problems, J. Algorithms 29(1)
(1998) 174–200.
[17] S. Khot, On the power of unique 2-prover 1-round games, in Proc.34th Annual ACM
Symp. Theory of Computing (Montreal Quebec, Canada, 2002), pp. 767–775.
[18] S. Khot and O. Regev, Vertex cover might be hard to approximate to within 2 − ,
J. Comput. Syst. Sci. 74(3) (2008) 335–349.
[19] R. Krithika, A. Rai, S. Saurabh and P. Tale, Parameterized and exact algorithms for
class domination coloring, in Int. Conf. Current Trends in Theory and Practice of
Informatics (Springer, 2017), pp. 336–349.
[20] C. Lund and M. Yannakakis, On the hardness of approximating minimization prob-
lems, J. ACM 41 (1994) 960–981.

2250188-23
July 31, 2023 11:18 WSPC/S1793-8309 257-DMAA 2250188

S. Das & S. Mishra

[21] H. B. Merouane, M. Haddad, M. Chellali and H. Kheddouci, Dominated colorings of


graphs, Graphs Combin. 31(3) (2015) 713–727.
[22] G. L. Nemhauser and L. E. Trotter, Vertex packings: structural properties and algo-
rithms, Math. Programming 8(1) (1975) 232–248.
[23] M. Shalu, S. Vijayakumar and T. Sandhya, On the complexity of cd-coloring of
graphs, Discrete Appl. Math. 280 (2020) 171–185.
[24] V. Swaminathan and R. Sundareswaran, Color class domination in graphs, in Math-
ematical and Experimental Physics (Narosa Publishing House, 2010).
[25] Y. Zhou and D. Zhao, On domination coloring in graphs, preprint (2019),
arXiv:1909.03715.
[26] D. Zuckerman, Linear degree extractors and the inapproximability of max clique and
chromatic number, in Proc. 38th Annual ACM Symp. Theory of Computing (2006),
pp. 681–690.

2250188-24

You might also like