Information Processing Letters 142 (2019) 20–26
Contents lists available at ScienceDirect
Information Processing Letters
www.elsevier.com/locate/ipl
A (3 + )k-vertex kernel for edge-disjoint triangle packing
Weibo Lin, Mingyu Xiao
∗
School of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu, China
a r t i c l e i n f o a b s t r a c t
Article history:
Received
14 July 2017
Received
in revised form 6 October 2018
Accepted
9 October 2018
Available
online 11 October 2018
Communicated
by Marcin Pilipczuk
Keywords:
Kernelization
Triangle
packing
Fixed-parameter
tractable
Graph
algorithms
In the Edge-Disjoint Triangle Packing problem (Etp), we are given a graph G and an
integer k, and asked to check whether G contains at least k edge-disjoint triangles. This
problem is NP-hard and has been well studied in the parameterized complexity. A vertex
kernel of size 4k was proved many years ago. Recently, the size of the vertex kernel was
improved to 3.5k. In this paper, we further improve the kernel size. We show that for any
fixed > 0, there is a polynomial-time algorithm that can reduce the input instance to an
equivalent instance of at most (3 + )k vertices.
© 2018 Elsevier B.V. All rights reserved.
1. Introduction
Kernelization, a preprocessing technique originating
from the field of parameterized algorithms [1], nowadays
plays an important role in computational complexity and
algorithms. A kernelization for a parameterized problem Q
is
an algorithm that, given an instance (I, k) of Q , works in
polynomial time and returns an equivalent instance (I
, k
)
of Q such that (I, k) is a yes-instance if and only if (I
, k
)
is a yes-instance, where k
≤ k and |I
| ≤ g(k) for some
computable function g only of k. The new instance (I
, k
)
is called a kernel of Q and the function g(·) is the size of
the kernel. If the upper bound g(·) is a polynomial (linear)
function of the parameter, then we say that Q admits a
polynomial (linear) kernel.
Initially,
kernelization was a technique introduced to
prove the fixed-parameter tractability of parameterized
problems. It is well known that a parameterized problem
is fixed-parameter tractable (FPT) if and only if it admits a
kernelization [2]. Moreover, a fixed-parameter tractable al-
gorithm
usually can be modified to get an exponential-size
kernel of the same problem directly [3]. A more interesting
*
Corresponding author.
E-mail
address: myxiao@gmail.com (M. Xiao).
problem is to find out whether an FPT problem has poly-
nomial
kernels or even linear kernels. Kernelization also
draws certain interests in practical applications, since it is
a powerful tool to evaluate the performance of some pre-
processing
algorithms.
For
most important NP-hard problems, kernelization
has been very well studied and small kernels for them
have been found. For example, a 2k-vertex kernel for the
Vertex
Cover problem [4,5], a 2k-vertex kernel for the
Cluster
Editing problem [6], a 4k
2
-vertex kernel for the
Feedback
Vertex Set problem [7], O (kd)-vertex kernels for
the d-Degree Bounded Vertex Deletion problem [8] and
the d-Size Separator problem [9], linear vertex kernels for
several planar graph problems [10], and so on.
In
this paper, we consider the Edge-Disjoint Trian-
gle
Packing problem (Etp). It is to check whether a
given graph contains at least k edge-disjoint triangles.
Etp is
NP-hard, even in planar graphs with maximum
degree 5 [11]. Etp is known to be APX-hard in general
graphs [12]. A general result of [13]leads to a polynomial-
time
(3/2 + )-approximation algorithm for any constant
> 0. When the graphs are restricted to planar graphs, the
result can be improved. Baker [14]gave a polynomial-time
approximation scheme for the vertex-disjoint problem ver-
sion
on planar graphs, which can be extended to Etp
on
planar graphs. In terms of parameterized complexity,
https://doi.org/10.1016/j.ipl.2018.10.006
0020-0190/
© 2018 Elsevier B.V. All rights reserved.