A Note on Max $k$-Vertex Cover: Faster FPT-AS, Smaller Approximate Kernel and Improved Approximation

TL;DR

This paper introduces faster fixed-parameter tractable approximation schemes and smaller kernels for Max and Min k-Vertex Cover, improving computational efficiency and approximation quality over previous methods.

Contribution

It presents a simplified FPT approximation scheme, an efficient approximate kernelization, and applies these to improve approximation algorithms for Max and Min k-Vertex Cover.

Findings

FPT-AS for Max k-VC runs in (1/ε)^{O(k)} poly(n) time, faster than previous.

Achieves an O(k/ε)-vertex approximate kernel for Max k-VC, smaller than prior unweighted kernels.

Provides a 0.92-approximation for Max k-VC using the kernel and SDP-based algorithms.

Abstract

In Maximum $k$-Vertex Cover (Max $k$-VC), the input is an edge-weighted graph $G$ and an integer $k$, and the goal is to find a subset $S$ of $k$ vertices that maximizes the total weight of edges covered by $S$. Here we say that an edge is covered by $S$ iff at least one of its endpoints lies in $S$. We present an FPT approximation scheme (FPT-AS) that runs in $(1/\epsilon)^{O(k)} poly(n)$ time for the problem, which improves upon Gupta et al.'s $(k/\epsilon)^{O(k)} poly(n)$-time FPT-AS [SODA'18, FOCS'18]. Our algorithm is simple: just use brute force to find the best $k$-vertex subset among the $O(k/\epsilon)$ vertices with maximum weighted degrees. Our algorithm naturally yields an efficient approximate kernelization scheme of $O(k/\epsilon)$ vertices; previously, an $O(k^5/\epsilon^2)$-vertex approximate kernel is only known for the unweighted version of Max $k$-VC [Lokshtanov et al., STOC'17]. Interestingly, this has an application outside of parameterized complexity: using our approximate kernelization as a preprocessing step, we can directly apply Raghavendra and Tan's SDP-based algorithm for 2SAT with cardinality constraint [SODA'12] to give an $0.92$-approximation for Max $k$-VC in polynomial time. This improves upon Feige and Langberg's algorithm [J. Algorithms'01] which yields $(0.75 + \delta)$-approximation for some (unspecified) constant $\delta > 0$. We also consider the minimization version (Min $k$-VC), where the goal is to minimize the total weight of edges covered by $S$. We provide an FPT-AS for Min $k$-VC with similar running time of $(1/\epsilon)^{O(k)} poly(n)$, which again improves on a $(k/\epsilon)^{O(k)} poly(n)$-time FPT-AS of Gupta et al. On the other hand, we show that there is unlikely a polynomial size approximate kernelization for Min $k$-VC for any factor less than two.