Searcharxiv⌕ Search

arXiv subjects

Guy Kortsarz

Publications and source records attributed to Guy Kortsarz.

28 records · Page 2Linked to original sources
arXiv2015

A $1.75$ LP approximation for the Tree Augmentation Problem

In the Tree Augmentation Problem (TAP) the goal is to augment a tree $T$ by a minimum size edge set $F$ from a given edge set $E$ such that $T \cup F$ is $2$-edge-connected. The best approximation ratio known for TAP is $1.5$. In the more general Weighted TAP problem, $F$ should be of minimum weight. Weighted TAP admits several $2$-approximation algorithms w.r.t. to the standard cut LP-relaxation, but for all of them the performance ratio of $2$ is tight even for TAP. The problem is equivalent to the problem of covering a laminar set family. Laminar set families play an important role in the design of approximation algorithms for connectivity network design problems. In fact, Weighted TAP is the simplest connectivity network design problem for which a ratio better than $2$ is not known. Improving this "natural" ratio is a major open problem, which may have implications on many other network design problems. It seems that achieving this goal requires finding an LP-relaxation with integrality gap better than $2$, which is a long time open problem even for TAP. In this paper we introduce such an LP-relaxation and give an algorithm that computes a feasible solution for TAP of size at most $1.75$ times the optimal LP value. This gives some hope to break the ratio $2$ for the weighted case. Our algorithm computes some initial edge set by solving a partial system of constraints that form the integral edge-cover polytope, and then applies local search on $3$-leaf subtrees to exchange some of the edges and to add additional edges. Thus we do not need to solve the LP, and the algorithm runs roughly in time required to find a minimum weight edge-cover in a general graph.

cs.DS↗
arXiv2015

Approximation Algorithms for Connected Maximum Cut and Related Problems

An instance of the Connected Maximum Cut problem consists of an undirected graph G = (V, E) and the goal is to find a subset of vertices S $\subseteq$ V that maximizes the number of edges in the cut δ(S) such that the induced graph G[S] is connected. We present the first non-trivial Ω(1/log n) approximation algorithm for the connected maximum cut problem in general graphs using novel techniques. We then extend our algorithm to an edge weighted case and obtain a poly-logarithmic approximation algorithm. Interestingly, in stark contrast to the classical max-cut problem, we show that the connected maximum cut problem remains NP-hard even on unweighted, planar graphs. On the positive side, we obtain a polynomial time approximation scheme for the connected maximum cut problem on planar graphs and more generally on graphs with bounded genus.

cs.DS↗
arXiv2014

Approximating Source Location and Star Survivable Network Problems

In Source Location (SL) problems the goal is to select a mini-mum cost source set $S \subseteq V$ such that the connectivity (or flow) $ψ(S,v)$ from $S$ to any node $v$ is at least the demand $d_v$ of $v$. In many SL problems $ψ(S,v)=d_v$ if $v \in S$, namely, the demand of nodes selected to $S$ is completely satisfied. In a node-connectivity variant suggested recently by Fukunaga, every node $v$ gets a "bonus" $p_v \leq d_v$ if it is selected to $S$. Fukunaga showed that for undirected graphs one can achieve ratio $O(k \ln k)$ for his variant, where $k=\max_{v \in V}d_v$ is the maximum demand. We improve this by achieving ratio $\min\{p^*\lnk,k\}\cdot O(\ln (k/q^*))$ for a more general version with node capacities, where $p^*=\max_{v \in V} p_v$ is the maximum bonus and $q^*=\min_{v \in V} q_v$ is the minimum capacity. In particular, for the most natural case $p^*=1$ considered by Fukunaga, we improve the ratio from $O(k \ln k)$ to $O(\ln^2k)$. We also get ratio $O(k)$ for the edge-connectivity version, for which no ratio that depends on $k$ only was known before. To derive these results, we consider a particular case of the Survivable Network (SN) problem when all edges of positive cost form a star. We give ratio $O(\min\{\ln n,\ln^2 k\})$ for this variant, improving over the best ratio known for the general case $O(k^3 \ln n)$ of Chuzhoy and Khanna.

cs.DS↗
arXiv2013

Fixed-Parameter and Approximation Algorithms: A New Look

A Fixed-Parameter Tractable (\FPT) $ρ$-approximation algorithm for a minimization (resp. maximization) parameterized problem $P$ is an FPT algorithm that, given an instance $(x, k)\in P$ computes a solution of cost at most $k \cdot ρ(k)$ (resp. $k/ρ(k)$) if a solution of cost at most (resp. at least) $k$ exists; otherwise the output can be arbitrary. For well-known intractable problems such as the W[1]-hard {Clique} and W[2]-hard {Set Cover} problems, the natural question is whether we can get any \FPT-approximation. It is widely believed that both {Clique} and {Set-Cover} admit no FPT $ρ$-approximation algorithm, for any increasing function $ρ$. Assuming standard conjectures such as the Exponential Time Hypothesis (ETH) \cite{eth-paturi} and the Projection Games Conjecture (PGC) \cite{r3}, we make the first progress towards proving this conjecture by showing that 1. Under the ETH and PGC, there exist constants $F_1, F_2 >0$ such that the {Set Cover} problem does not admit an FPT approximation algorithm with ratio $k^{F_1}$ in $2^{k^{F_2}}\cdot \text{poly}(N,M)$ time, where $N$ is the size of the universe and $M$ is the number of sets. 2. Unless $\NP\subseteq \SUBEXP$, for every $1> δ> 0$ there exists a constant $F(δ)>0$ such that {Clique} has no FPT cost approximation with ratio $k^{1-δ}$ in $2^{k^{F}}\cdot \text{poly}(n)$ time, where $n$ is the number of vertices in the graph. In the second part of the paper we consider various W[1]-hard problems such as {\dst}, {\dsf}, Directed Steiner Network and {\mec}. For all these problem we give polynomial time $f(\text{OPT})$-approximation algorithms for some small function $f$ (the largest approximation ratio we give is $\text{OPT}^2$).

cs.DS↗
arXiv2013

Approximation Algorithms for Movement Repairmen

In the {\em Movement Repairmen (MR)} problem we are given a metric space $(V, d)$ along with a set $R$ of $k$ repairmen $r_1, r_2, ..., r_k$ with their start depots $s_1, s_2, ..., s_k \in V$ and speeds $v_1, v_2, ..., v_k \geq 0$ respectively and a set $C$ of $m$ clients $c_1, c_2, ..., c_m$ having start locations $s'_1, s'_2, ..., s'_m \in V$ and speeds $v'_1, v'_2, ..., v'_m \geq 0$ respectively. If $t$ is the earliest time a client $c_j$ is collocated with any repairman (say, $r_i$) at a node $u$, we say that the client is served by $r_i$ at $u$ and that its latency is $t$. The objective in the (\smr{}) problem is to plan the movements for all repairmen and clients to minimize the sum (average) of the clients latencies. The motivation for this problem comes, for example, from Amazon Locker Delivery \cite{amazon} and USPS gopost \cite{gopost}. We give the first $O(\log n)$-approximation algorithm for the \smr{} problem.

cs.DS↗
arXiv2012

Matroid Secretary for Regular and Decomposable Matroids

In the matroid secretary problem we are given a stream of elements and asked to choose a set of elements that maximizes the total value of the set, subject to being an independent set of a matroid given in advance. The difficulty comes from the assumption that decisions are irrevocable: if we choose to accept an element when it is presented by the stream then we can never get rid of it, and if we choose not to accept it then we cannot later add it. Babaioff, Immorlica, and Kleinberg [SODA 2007] introduced this problem, gave O(1)-competitive algorithms for certain classes of matroids, and conjectured that every matroid admits an O(1)-competitive algorithm. However, most matroids that are known to admit an O(1)-competitive algorithm can be easily represented using graphs (e.g. graphic and transversal matroids). In particular, there is very little known about F-representable matroids (the class of matroids that can be represented as elements of a vector space over a field F), which are one of the foundational matroid classes. Moreover, most of the known techniques are as dependent on graph theory as they are on matroid theory. We go beyond graphs by giving an O(1)-competitive algorithm for regular matroids (the class of matroids that are representable over every field), and use techniques that are matroid-theoretic rather than graph-theoretic. We use the regular matroid decomposition theorem of Seymour to decompose any regular matroid into matroids which are either graphic, cographic, or isomorphic to R_{10}, and then show how to combine algorithms for these basic classes into an algorithm for regular matroids. This allows us to generalize beyond regular matroids to any class of matroids that admits such a decomposition into classes for which we already have good algorithms. In particular, we give an O(1)-competitive algorithm for the class of max-flow min-cut matroids.

cs.DS↗
arXiv2012

Steiner Forest Orientation Problems

We consider connectivity problems with orientation constraints. Given a directed graph $D$ and a collection of ordered node pairs $P$ let $P[D]=\{(u,v) \in P: D {contains a} uv{-path}}$. In the {\sf Steiner Forest Orientation} problem we are given an undirected graph $G=(V,E)$ with edge-costs and a set $P \subseteq V \times V$ of ordered node pairs. The goal is to find a minimum-cost subgraph $H$ of $G$ and an orientation $D$ of $H$ such that $P[D]=P$. We give a 4-approximation algorithm for this problem. In the {\sf Maximum Pairs Orientation} problem we are given a graph $G$ and a multi-collection of ordered node pairs $P$ on $V$. The goal is to find an orientation $D$ of $G$ such that $|P[D]|$ is maximum. Generalizing the result of Arkin and Hassin [DAM'02] for $|P|=2$, we will show that for a mixed graph $G$ (that may have both directed and undirected edges), one can decide in $n^{O(|P|)}$ time whether $G$ has an orientation $D$ with $P[D]=P$ (for undirected graphs this problem admits a polynomial time algorithm for any $P$, but it is NP-complete on mixed graphs). For undirected graphs, we will show that one can decide whether $G$ admits an orientation $D$ with $|P[D]| \geq k$ in $O(n+m)+2^{O(k\cdot \log \log k)}$ time; hence this decision problem is fixed-parameter tractable, which answers an open question from Dorn et al. [AMB'11]. We also show that {\sf Maximum Pairs Orientation} admits ratio $O(\log |P|/\log\log |P|)$, which is better than the ratio $O(\log n/\log\log n)$ of Gamzu et al. [WABI'10] when $|P|<n$. Finally, we show that the following node-connectivity problem can be solved in polynomial time: given a graph $G=(V,E)$ with edge-costs, $s,t \in V$, and an integer $\ell$, find a min-cost subgraph $H$ of $G$ with an orientation $D$ such that $D$ contains $\ell$ internally-disjoint $st$-paths, and $\ell$ internally-disjoint $ts$-paths.

cs.DS↗
arXiv2012

Label Cover instances with large girth and the hardness of approximating basic k-spanner

We study the well-known Label Cover problem under the additional requirement that problem instances have large girth. We show that if the girth is some $k$, the problem is roughly $2^{\log^{1-ε} n/k}$ hard to approximate for all constant $ε> 0$. A similar theorem was claimed by Elkin and Peleg [ICALP 2000], but their proof was later found to have a fundamental error. We use the new proof to show inapproximability for the basic $k$-spanner problem, which is both the simplest problem in graph spanners and one of the few for which super-logarithmic hardness was not known. Assuming $NP \not\subseteq BPTIME(2^{polylog(n)})$, we show that for every $k \geq 3$ and every constant $ε> 0$ it is hard to approximate the basic $k$-spanner problem within a factor better than $2^{(\log^{1-ε} n) / k}$ (for large enough $n$). A similar hardness for basic $k$-spanner was claimed by Elkin and Peleg [ICALP 2000], but the error in their analysis of Label Cover made this proof fail as well. Thus for the problem of Label Cover with large girth we give the first non-trivial lower bound. For the basic $k$-spanner problem we improve the previous best lower bound of $Ω(\log n)/k$ by Kortsarz [Algorithmica 1998]. Our main technique is subsampling the edges of 2-query PCPs, which allows us to reduce the degree of a PCP to be essentially equal to the soundness desired. This turns out to be enough to essentially guarantee large girth.

cs.DS↗
arXiv2011

Combinatorial Algorithms for Capacitated Network Design

We focus on designing combinatorial algorithms for the Capacitated Network Design problem (Cap-SNDP). The Cap-SNDP is the problem of satisfying connectivity requirements when edges have costs and hard capacities. We begin by showing that the Group Steiner tree problem (GST) is a special case of Cap-SNDP even when there is connectivity requirement between only one source-sink pair. This implies the first poly-logarithmic lower bound for the Cap-SNDP. We next provide combinatorial algorithms for several special cases of this problem. The Cap-SNDP is equivalent to its special case when every edge has either zero cost or infinite capacity. We consider a special case, called Connected Cap-SNDP, where all infinite-capacity edges in the solution are required to form a connected component containing the sinks. This problem is motivated by its similarity to the Connected Facility Location problem [G+01,SW04]. We solve this problem by reducing it to Submodular tree cover problem, which is a common generalization of Connected Cap-SNDP and Group Steiner tree problem. We generalize the recursive greedy algorithm [CEK] achieving a poly-logarithmic approximation algorithm for Submodular tree cover problem. This result is interesting in its own right and gives the first poly-logarithmic approximation algorithms for Connected hard capacities set multi-cover and Connected source location. We then study another special case of Cap-SNDP called Unbalanced point-to-point connection problem. Besides its practical applications to shift design problems [EKS], it generalizes many problems such as k-MST, Steiner Forest and Point-to-Point Connection. We give a combinatorial logarithmic approximation algorithm for this problem by reducing it to degree-bounded SNDP.

cs.DS↗