See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/282375697

Vertex Cover Problem—Revised Approximation Algorithm

Article in Advances in Intelligent Systems and Computing · November 2015

DOI: 10.1007/978-81-322-2126-5_2

CITATIONS READS

4 2,378

3 authors, including:

Selvakumar Ramachandran
VIT University
58 PUBLICATIONS 112 CITATIONS

SEE PROFILE

All content following this page was uploaded by Selvakumar Ramachandran on 29 June 2020.

The user has requested enhancement of the downloaded file.

Vertex Cover Problem—Revised
Approximation Algorithm

Kartik Shah, Praveenkumar Reddy and R. Selvakumar

Abstract This paper is aimed to present the solution to vertex cover problem by
means of an approximation solution. As it is NP complete problem, we can have an
approximate time algorithm to solve the vertex cover problem. We will modify the
algorithm to have an algorithm which can be solved in polynomial time and which
will give near to optimum solution. It is a simple algorithm which will be based on
articulation point. Articulation point can be found using the Depth First Search
algorithm.

Keywords Articulation point
Vertex covering problem
Optimization

Approximation algorithm

1 Introduction

A problem (P) is called NP complete if P is in class NP and every problem P′ is

polynomially reducible to P. Vertex cover problem is that problem in which we
take an undirected graph (G = (V, E)) with ‘V’ vertices and ‘E’ edges and will have
a set which contains the vertices which can cover all the edges of the graph.
According to Garey and Johnson, vertex cover problem is one of the six basic NP
complete problems [1].
It is observed that sometimes, there exist some problems for which we do not
have an optimal solutions. However, we can have an approximate solution which

K. Shah (&)
P. Reddy
R. Selvakumar

School of Computing Science and Engineering, VIT University, Vellore 632014, India
e-mail: kartikshah@mail.com
P. Reddy
e-mail: praveenkumar.reddym2012@vit.ac.in
R. Selvakumar
e-mail: rselvakumar@vit.ac.in

© Springer India 2015 9

L.P. Suresh et al. (eds.), Artificial Intelligence and Evolutionary Algorithms
in Engineering Systems, Advances in Intelligent Systems and Computing 324,
DOI 10.1007/978-81-322-2126-5_2
10 K. Shah et al.

will be closer to the optimal solution. The algorithm which provides closer solution
to the optimal solution is known as approximation algorithms.
Let G = (V, E) be a graph where V is number of vertices and E be number of
edges. The set of vertices V 0  V is said to be cover if for each edge ðu; vÞ 2 E,
either u 2 V 0 or v 2 V 0 or both. The number of vertices in V′ is known as the size of
the cover V′. The problem of finding minimum size is known as vertex cover
problem.

2 Related Work

For vertex cover problems, we have many real-life problems in which vertex cover
problem solution can be applied. Like, to find population growths taken into
polynomial time and for that we can take bi-parted graph and also take articulation
points [2]. Also, vertex cover problem for network base routing delays in tolerance
network [3], for network traffic measurement [4]. Polynomial space parameterized
vertex cover can be solved in O(1.2738k + kn) time [5]. F. Delbot and C. Laforest
had proposed an algorithm in which vertexes are scanned from left to right. Con-
dition is as follows: “u is added to vertex cover if and only if it has at least one
neighbor not in the cover” [6]. This algorithm is called as LIST LEFT which is
given as follows [7]:

Labeled graph L(G) = (L(V), E)

1. C /
2. For each vertex, u 2 LðVÞ do
3. If u has at least one right neighbor, then
4. C C [ fug
5. Return C
Sorted LL is another technique of finding vertex cover. It takes decision as “if
there is at least one v 2 NðuÞ with lower degree, select u; otherwise, if u has only
neighbor with higher degree, u is not selected.” Algorithm can be represented as
follows [7]:
Labeled graph L(G) = (L(V), E)
1. C /
2. For each vertex, u 2 LðVÞ do
3. If u has at least one right neighbor with a lower degree or a right neighbor with
the same degree, then
4. C C [ fug
5. Return C
One more approach is to use ANTI SORTED LL. In this algorithm, degree of
vertex’s neighbor is taken into consider. Below is the algorithm shown [7]:
Vertex Cover Problem—Revised Approximation Algorithm 11

Labeled graph L(G) = (L(V), E)

1. C /
2. For each vertex, u 2 LðVÞ do
3. If u has at least one right neighbor with a larger degree or a left neighbor with
the same degree, then
4. C C [ fug
5. Return C
These all algorithms can be used for finding vertex cover set of a graph.
It is possible to approximate the weight of the vertices which results in local
approximation and called as local-ratio theorem and is so called local-ratio theorem for
weighted vertex cover problem [8]. Further, it states that it is possible to put together the
Nemhauser–Trotter algorithm (local optimization algorithm) and local-ratio theorem to
get new approximation techniques that improve performance [8]. Kernelization is the
process of applying polynomial time preprocessing to the instance of graph G(V, E)
and obtaining other instance G′(V′, E′) where V′ ≤ V and G′ will have vertices of vertex
cover V′ [9]. It is observed that we can find up to 3/2 vertices of minimum vertex covers
using collection of graph transformations. The algorithm guarantees an approximation
ratio of 3/2, for finding large number of randomly created graphs. The reductions are
extremely fast. The problem has best-case and worst-case approximation ratio as 2-O
(1) [10]. The random graphs that we generally use are simple random graph model
proposed by y Erdos and Renyi [11]. Chromatic number is defined as minimum number
of colors used for vertex to cover all with distance 1. Chromatic number can be used to
find the vertex over of the graph. Kuhn and Mastrolilli [12] investigated weighted
vertex cover problem for graphs when a locally bounded coloring is given.

3 Approach

There are many approaches to solve vertex cover problem.

3.1 Classical Approach

The approximation algorithm which is available to solve the vertex cover problem
is a polynomial time algorithm, and it can be solved in O (V + E) time complexity
where V is number of vertices and E is number of edges. The approximation vertex
cover is a polynomial time 2 approximation algorithm.
That is, let V be a vertex cover and V* be the optimal cover to that problem.
Then, we can prove that

jV j  2jV  j
12 K. Shah et al.

So, in this case, if number of vertices increases, the solution to vertex cover may
also diverse far from optimal solution.

Algorithm 1.1 Approximation Algorithm [13]

APPROX VERTEX COVER(G)
1. C /
2. E′ = E[G]
3. While E 0 /
4. Let (u, v) be an arbitrary edge of E′
5. C C [ fu; vg
6. Remove from E′ every edge incident on either u or v
7. Return C

3.2 Our Approach

As we noted above that approximate vertex cover problem solution is less than or
equal to 2 times the optimal solution. We proposed the algorithm which will take
near to optimal solution.
In our approach, we will find the articulation point of that particular graph and
then add those vertices in the vertex cover. From remaining edges, we will take the
common vertex which can cover 2 or more edges. After that if some more edges are
not covered, then we will take 1 of the vertex from the edge. In order to find the
articulation point, we can go for Depth First Search (DFS) algorithm, which takes
O (V + E) time to find the articulation points. There are many more algorithms
available for finding articulation points. DFS is one of the simplest algorithms.
Proposed algorithm to find the vertex cover of graph is shown below.

Algorithm 1.2 Proposed Vertex Cover Algorithm

1. C /
2. C′ = ArticPointDFS (vertex);
3. C = ←C U C′
4. Repeat Step 2 and Step 3 until all the Articulation Point is found;
5. E′ = E [G];
6. E′ = E′ − {set of edges covered by C′}
7. While E 0 6¼ /
Let (a, b) be an arbitrary edge, check (a, b) ∩ {take each edge of E′} ≠ ϕ Then,
add any of vertices ‘a’ or ‘b’ in C.
Remove edge (a, b) from E′
Else add the common vertex.
Remove edges containing that vertices from E′
8. Return C.
Vertex Cover Problem—Revised Approximation Algorithm 13

4 Analysis

Now comparing both the algorithms, let us take 1 case of finding vertex cover.
Consider the following example. It contains 10 vertices and 13 edges. All vertices
are connected with edges as shown in Fig. 1. Now, we will apply the vertex cover
algorithms to the following example with both the approaches (i.e., existing
approach and proposed approach). After that result will be compared for example,
and output from each algorithm will be compared. The algorithm is computed step
by step as per the algorithms discussed above.
Using existing algorithm:
1. C /
2. E′ = {(1, 2), (2, 3), (3, 4), (1, 4), (2, 7), (2, 9), (2, 10), (3, 5), (3, 6), (7, 8), (7, 9),
(7, 10), (9, 10)}
3. Let us take an arbitrary edge (9, 10)
4. C = {9, 10}
5. E′ = {(1, 2), (2, 3), (3, 4), (1, 4), (2, 7), (2, 9), (2, 10), (3, 5), (3, 6), (7, 8)}
6. Let us take another arbitrary edge (1, 4)
7. C = {1, 4, 9, 10}
8. E′ = {(2, 3), (2, 7), (2, 9), (2, 10), (3, 5), (3, 6), (7, 8)}

Fig. 1 Vertices are connected

with each other, and a 8
complex graph is formed

9
1 2

10

4 3

6 5
14 K. Shah et al.

9. Let us take another arbitrary edge (3, 5)

10. C = {1, 3, 4, 5, 9, 10}
11. E′ = {(2, 7), (7, 8)}
12. Let us take another arbitrary edge (7, 8)
13. C = {1, 3, 4, 5, 7, 8, 9, 10}
Finally, vertex cover contains 8 elements.
Using proposed algorithm:
1. C /
2. E′ = {(1, 2), (2, 3), (3, 4), (1, 4), (2, 7), (2, 9), (2, 10), (3, 5), (3, 6), (7, 8), (7, 9),
(7, 10), (9, 10)}
3. Find articulation points, C′ = {2, 3, 7}
4. C = {2, 3, 7}
5. E′ = E′ − {set of edges covered by C′}
6. E′ = {(1, 4), (9, 10)}
7. Take (1, 4) as an arbitrary edge.
8. C = {1, 2, 3, 7}
9. E′ = {(9, 10)}
10. Take (9, 10) as an arbitrary edge.
11. E 0 ¼ /
12. C = {1, 2, 3, 7, 9}
Finally, vertex cover contains 5 elements.
As we can see the result from existing approach and from proposed approach, it
is clear that the existing approach can give us the true result but that will not be
minimal. Some vertices that are not needed will also be included in the vertex cover
list, while the proposed approach will have less number of vertices in the vertex
cover list and which covers all the edges. Using this approach, first we will collect
all the articulation point details and will include that vertices into vertex cover list.
After that we will go with normal approach for remaining vertices, and the result
will be as shown in above example. The implementation of the above approach is
discussed in following section.

5 Implementation

Implementation of proposed algorithm can be done in 2 steps:

Step 1: Find the articulation points of the graph given using articulation point
algorithm (Using DFS algorithm). Add articulation points in the set of
vertex cover. For example, given above, after applying algorithm, we can
get following output (Fig. 2).
Vertex Cover Problem—Revised Approximation Algorithm 15

Fig. 2 Step 1 of
implementation

Step 2: Remove the edges which are adjacent to all the vertices listed in Step 1.
Apply approx vertex cover algorithm on remaining graph (Disconnected
Graph). For example, given above, we have 4 choices {(1, 9) or (1, 10) or
(4, 9) or (4, 10)}. We can choose any combination.
Step 3: Take union of vertices found in Steps 1 and 2.

6 Conclusion

From above example, we can see that using existing algorithm, we are getting 8
elements in vertex cover set. While using proposed algorithm, we are getting 5
elements. Also, it is same as minimal vertex cover.
If V is a vertex cover set derived from Algorithm 1.1 and V* is an optimal vertex
cover set, then

jV j  2jV  j

If C is a vertex cover set derived from Algorithm 1.3 and C* is an optimal vertex
cover set, then

jC j  jC  j

Proposed algorithm takes O (2 (V + E)) time complexity. O (V + E) for DFS and

O (V + E) for finding vertex cover. So, total O (V + E) time to compute vertex cover
is same as the available algorithm, but our algorithm provides much nearer solution
to optimal solution.
16 K. Shah et al.

7 Future Work

In this proposed algorithm, we are using DFS algorithm to find the articulation
points of the graph. This algorithm takes O (V + E) time to compute the articulation
points. We can go for some other techniques to find the articulation points which
can take less time compared to DFS. Also, we can find some other techniques
which always provide the exact solution to optimal solution. Also, some more
techniques can be used to provide nearer solution. There is also possible to find
algorithm which runs faster than this algorithm.

Acknowledgment We would like to thank the School of Computer Science and Engineering,
VIT University, for giving us such an opportunity to carry out this research work and also for
providing us the requisite resources and infrastructure for carrying out the research.

References

1. M. Garry, D. Johnson, Computers and Intractability: A User Guide to the Theory of NP

Completeness (San Francisco, 1979)
2. P.S. Oliveto, X. Yao, J. He, Analysis of Population-based Evolutionary Algorithms for the
Vertex Cover Problem (IEEE, 2008), pp. 1563–1570
3. L. Ding, B. Gu, X. Hong, B. Dixon, Articulation node based routing in delay tolerant
networks, in IEEE International Conference (2009), pp. 1–6
4. Y. Zeng, D. Wang, W. Liu, A. Xiong, An approximation algorithm for weak vertex cover
problem in IP network traffic measurement, IEEE International Conference (2009),
pp. 182–186
5. J. Chen, I.A. Kanj, G. Xia, Improved parameterized upper bounds for vertex cover. 31st
International Conference on Mathematical Foundations of Computer Science (2006)
6. F. Delbot, C. Laforest, A better list heuristic for vertex cover. Inf. Process. Lett. 107, 125–127
(2008)
7. E. Angel, R. Campigotto, C. Laforest, Algorithm for the vertex cover problem on large graphs.
IBISC Research report (2010)
8. R. Bar-Yehuda, S. Even, A local-ratio theorem for approximating the weighted vertex cover
problem. Ann. Discrete Math. 25, 27–46 (1985)
9. G.L. Nemhauser, L.E. Trotter, Vertex packing: structural properties and algorithms. Math.
Program. 8, 232–248 (1975)
10. E. Asgeirsson, C. Stein, Vertex Cover Approximations on Random Graphs (Springer, Berlin,
2007), pp. 285–296
11. P. Erdos, A. Renyi, On random graphs. Publ. Math. Debrecen 6, 290–297 (1959)
12. F. Kuhn, M. Mastrolilli, Vertex cover in graphs with locally few colors (2011), pp 498–509
13. D. Hochbaum, Approximation algorithms for the set covering and vertex cover problems.
SIAM J. Comput. 11(3), 555–556 (1982)
http://www.springer.com/978-81-322-2125-8

View publication stats