arXiv subjects

Meirav Zehavi

Publications and source records attributed to Meirav Zehavi.

94 records · Page 6Linked to original sources

arXiv2015

Spotting Trees with Few Leaves

We show two results related to the Hamiltonicity and $k$-Path algorithms in undirected graphs by Björklund [FOCS'10], and Björklund et al., [arXiv'10]. First, we demonstrate that the technique used can be generalized to finding some $k$-vertex tree with $l$ leaves in an $n$-vertex undirected graph in $O^*(1.657^k2^{l/2})$ time. It can be applied as a subroutine to solve the $k$-Internal Spanning Tree ($k$-IST) problem in $O^*(\min(3.455^k, 1.946^n))$ time using polynomial space, improving upon previous algorithms for this problem. In particular, for the first time we break the natural barrier of $O^*(2^n)$. Second, we show that the iterated random bipartition employed by the algorithm can be improved whenever the host graph admits a vertex coloring with few colors; it can be an ordinary proper vertex coloring, a fractional vertex coloring, or a vector coloring. In effect, we show improved bounds for $k$-Path and Hamiltonicity in any graph of maximum degree $Δ=4,\ldots,12$ or with vector chromatic number at most 8.

cs.DS↗

arXiv2014

Representative Families: A Unified Tradeoff-Based Approach

Let $M=(E,{\cal I})$ be a matroid, and let $\cal S$ be a family of subsets of size $p$ of $E$. A subfamily $\widehat{\cal S}\subseteq{\cal S}$ represents ${\cal S}$ if for every pair of sets $X\in{\cal S}$ and $Y\subseteq E\setminus X$ such that $X\cup Y\in{\cal I}$, there is a set $\widehat{X}\in\widehat{\cal S}$ disjoint from $Y$ such that $\widehat{X}\cup Y\in{\cal I}$. Fomin et al. (Proc. ACM-SIAM Symposium on Discrete Algorithms, 2014) introduced a powerful technique for fast computation of representative families for uniform matroids. In this paper, we show that this technique leads to a unified approach for substantially improving the running times of parameterized algorithms for some classic problems. This includes, among others, $k$-Partial Cover, $k$-Internal Out-Branching, and Long Directed Cycle. Our approach exploits an interesting tradeoff between running time and the size of the representative families.

cs.DS↗

arXiv2014

Parameterized Algorithms for Graph Partitioning Problems

We study a broad class of graph partitioning problems, where each problem is specified by a graph $G=(V,E)$, and parameters $k$ and $p$. We seek a subset $U\subseteq V$ of size $k$, such that $α_1m_1 + α_2m_2$ is at most (or at least) $p$, where $α_1,α_2\in\mathbb{R}$ are constants defining the problem, and $m_1, m_2$ are the cardinalities of the edge sets having both endpoints, and exactly one endpoint, in $U$, respectively. This class of fixed cardinality graph partitioning problems (FGPP) encompasses Max $(k,n-k)$-Cut, Min $k$-Vertex Cover, $k$-Densest Subgraph, and $k$-Sparsest Subgraph. Our main result is an $O^*(4^{k+o(k)}Δ^k)$ algorithm for any problem in this class, where $Δ\geq 1$ is the maximum degree in the input graph. This resolves an open question posed by Bonnet et al. [IPEC 2013]. We obtain faster algorithms for certain subclasses of FGPPs, parameterized by $p$, or by $(k+p)$. In particular, we give an $O^*(4^{p+o(p)})$ time algorithm for Max $(k,n-k)$-Cut, thus improving significantly the best known $O^*(p^p)$ time algorithm.

cs.DS↗

arXiv2013

Deterministic Parameterized Algorithms for Matching and Packing Problems

We present three deterministic parameterized algorithms for well-studied packing and matching problems, namely, Weighted q-Dimensional p-Matching ((q,p)-WDM) and Weighted q-Set p-Packing ((q,p)-WSP). More specifically, we present an O*(2.85043^{(q-1)p}) time deterministic algorithm for (q,p)-WDM, an O*(8.04143^p) time deterministic algorithm for the unweighted version of (3,p)-WDM, and an O*((0.56201\cdot 2.85043^q)^p) time deterministic algorithm for (q,p)-WSP. Our algorithms significantly improve the previously best known O* running times in solving (q,p)-WDM and (q,p)-WSP, and the previously best known deterministic O* running times in solving the unweighted versions of these problems. Moreover, we present kernels of size O(e^qq(p-1)^q) for (q,p)-WDM and (q,p)-WSP, improving the previously best known kernels of size O(q!q(p-1)^q) for these problems.

cs.DS↗