arXiv subjects

Oded Lachish

Publications and source records attributed to Oded Lachish.

24 records · Page 2Linked to original sources

arXiv2014

O(log log rank) Competitive-Ratio for the Matroid Secretary Problem

In the \textit{Matroid Secretary Problem} (MSP), the elements of the ground set of a Matroid are revealed on-line one by one, each together with its value. An algorithm for the MSP is \textit{Matroid-Unknown} if, at every stage of its execution: (i) it only knows the elements that have been revealed so far and their values, and (ii) it has access to an oracle for testing whether or not any subset of the elements that have been revealed so far is an independent set. An algorithm is \textit{Known-Cardinality} if, in addition to (i) and (ii), it also initially knows the cardinality of the ground set of the Matroid. We present here a Known-Cardinality and \textit{Order-Oblivious} algorithm that, with constant probability, selects an independent set of elements, whose value is at least the optimal value divided by $O(\log{\logρ})$, where $ρ$ is the rank of the Matroid; that is, the algorithm has a \textit{competitive-ratio} of $O(\log{\logρ})$. The best previous results for a Known-Cardinality algorithm are a competitive-ratio of $O(\logρ)$, by Babaioff \textit{et al.} (2007), and a competitive-ratio of $O(\sqrt{\logρ})$, by Chakraborty and Lachish (2012). In many non-trivial cases the algorithm we present has a competitive-ratio that is better than the $O(\log{\logρ})$. The cases in which it fails to do so are easily characterized. Understanding these cases may lead to improved algorithms for the problem or, conversely, to non-trivial lower bounds.

cs.DS↗

arXiv2014

Testing Formula Satisfaction

We study the query complexity of testing for properties defined by read once formulas, as instances of {\em massively parametrized properties}, and prove several testability and non-testability results. First we prove the testability of any property accepted by a Boolean read-once formula involving any bounded arity gates, with a number of queries exponential in $ε$, doubly exponential in the arity, and independent of all other parameters. When the gates are limited to being monotone, we prove that there is an {\em estimation} algorithm, that outputs an approximation of the distance of the input from satisfying the property. For formulas only involving And/Or gates, we provide a more efficient test whose query complexity is only quasipolynomial in $ε$. On the other hand, we show that such testability results do not hold in general for formulas over non-Boolean alphabets; specifically we construct a property defined by a read-once arity $2$ (non-Boolean) formula over an alphabet of size $4$, such that any $1/4$-test for it requires a number of queries depending on the formula size. We also present such a formula over an alphabet of size $5$ that additionally satisfies a strong monotonicity condition.

cs.DS↗

arXiv2013

Some properties are not even partially testable

For a property $P$ and a sub-property $P'$, we say that $P$ is $P'$-partially testable with $q$ queries if there exists an algorithm that distinguishes, with high probability, inputs in $P'$ from inputs $ε$-far from $P$ by using $q$ queries. There are natural properties that require many queries to test, but can be partitioned into a small number of subsets for which they are partially testable with very few queries. We prove the existence of a property $P$ such that the only subsets $P'$ for which $P$ is $P'$-partially testable are very small. To prove this we introduce new techniques for proving property testing lower bounds. In addition to obtaining some broad-brush criteria for non-testability, this implies a lower bound on the possibility of PCPPs with a sublinear proof. This also implies lower bounds on MAPs, a notion newly defined by Gur and Rothblum. The new techniques rely on analyzing a proposed partial tester. We show that the queries performed by a tester must, with high probability, query indexes where a uniformly random member of the sub-property has low entropy. We then show how one can aggregate the "entropy loss" to deduce that a random choice in the sub-property must have low entropy, and therefore the sub-property must be small. We develop two techniques for aggregating the entropy loss. A simpler technique that applies to non-adaptive testers is based on partitioning the input bits into high query probability parts and parts where there is an entropy loss when conditioned on the high probability parts. Against adaptive testers we develop a technique based on constructing a decision tree. The root-to-leaf paths in this tree rearrange the input into parts where each exhibits entropy loss when conditioned on the path prefix. This decision tree is constructed by combining carefully selected decision trees from those used by the adaptive testing algorithm.

cs.CC↗

arXiv2010

Two-phase algorithms for the parametric shortest path problem

A {\em parametric weighted graph} is a graph whose edges are labeled with continuous real functions of a single common variable. For any instantiation of the variable, one obtains a standard edge-weighted graph. Parametric weighted graph problems are generalizations of weighted graph problems, and arise in various natural scenarios. Parametric weighted graph algorithms consist of two phases. A {\em preprocessing phase} whose input is a parametric weighted graph, and whose output is a data structure, the advice, that is later used by the {\em instantiation phase}, where a specific value for the variable is given. The instantiation phase outputs the solution to the (standard) weighted graph problem that arises from the instantiation. The goal is to have the running time of the instantiation phase supersede the running time of any algorithm that solves the weighted graph problem from scratch, by taking advantage of the advice. In this paper we construct several parametric algorithms for the shortest path problem. For the case of linear function weights we present an algorithm for the single source shortest path problem.

cs.CC↗

arXiv2009

Wiretapping a hidden network

We consider the problem of maximizing the probability of hitting a strategically chosen hidden virtual network by placing a wiretap on a single link of a communication network. This can be seen as a two-player win-lose (zero-sum) game that we call the wiretap game. The value of this game is the greatest probability that the wiretapper can secure for hitting the virtual network. The value is shown to equal the reciprocal of the strength of the underlying graph. We efficiently compute a unique partition of the edges of the graph, called the prime-partition, and find the set of pure strategies of the hider that are best responses against every maxmin strategy of the wiretapper. Using these special pure strategies of the hider, which we call omni-connected-spanning-subgraphs, we define a partial order on the elements of the prime-partition. From the partial order, we obtain a linear number of simple two-variable inequalities that define the maxmin-polytope, and a characterization of its extreme points. Our definition of the partial order allows us to find all equilibrium strategies of the wiretapper that minimize the number of pure best responses of the hider. Among these strategies, we efficiently compute the unique strategy that maximizes the least punishment that the hider incurs for playing a pure strategy that is not a best response. Finally, we show that this unique strategy is the nucleolus of the recently studied simple cooperative spanning connectivity game.

cs.GT↗

arXiv2009

Spanning connectivity games

The Banzhaf index, Shapley-Shubik index and other voting power indices measure the importance of a player in a coalitional game. We consider a simple coalitional game called the spanning connectivity game (SCG) based on an undirected, unweighted multigraph, where edges are players. We examine the computational complexity of computing the voting power indices of edges in the SCG. It is shown that computing Banzhaf values and Shapley-Shubik indices is #P-complete for SCGs. Interestingly, Holler indices and Deegan-Packel indices can be computed in polynomial time. Among other results, it is proved that Banzhaf indices can be computed in polynomial time for graphs with bounded treewidth. It is also shown that for any reasonable representation of a simple game, a polynomial time algorithm to compute the Shapley-Shubik indices implies a polynomial time algorithm to compute the Banzhaf indices. As a corollary, computing the Shapley value is #P-complete for simple games represented by the set of minimal winning coalitions, Threshold Network Flow Games, Vertex Connectivity Games and Coalitional Skill Games.

cs.GT↗