Dans cette thèse on s’intéresse à un modèle de faute Byzantins Mobiles. Jusqu’à présent, seulement le problème du Consensus a été résolu en présente de faute Byzantines Mobiles et plusieurs variations de ce modèle de faute ont été proposé. Pour chacun de ces modelés ont été prouvées les bornes inferieures du nombre de processus correct nécessaire et des solutions asymptotiquement optimales ont été proposées. Notre première contribution porte sur les registres repartis dans ce modèle. Les registres repartis sont l’abstraction à la base du stockage reparti. Ces résultats préconisent donc notre deuxième contribution principale, un modèle de faute Byzantine Mobile généralisé. Notre troisième contribution est un ensemble de preuves de nécessité et d’impossibilité pour les registres repartis dans ces modèles. En particulier on prouve qu’il n’est pas possible d’implémenter la spécification plus faible des registres dans un système asynchrone. Par contre, pour les systèmes synchrones, on prouve des bornes inferieures et propose des protocoles asymptotiquement optimaux pour le registrer régulier. Pour conclure, notre dernière contribution porte sur le problème d’accord approximé, une forme affaiblie du consensus. On résout ce problème dans le modèle basé sur ronde, le même du consensus. En outre, il est intéressant de noter qua dans le modèle statique, la borne inferieure sur le nombre de répliques est la même pour le consensus et pour le problème d’accord approximé. Le même invariant s’applique avec les fautes byzantine mobiles. De plus, on accompagne ces bornes inferieures avec une solution asymptotiquement optimale pour le problème d’accord approximé. / In this thesis we consider a model where Byzantine failures are not fixed, we consider the so called Mobile Byzantine failures. So far, only Consensus problem has been solved in presence of Mobile Byzantine failures and interestingly different variations of this failure model have been proposed. For each of them have been proved lower bounds on the number of required processes and have been proposed tight solutions. Our first contribution concerns distributed Registers in such strong model. Distributed Registers are the basic abstraction for Distributed Storages. This advocates our second and main contribution, a general Mobile Byzantine Failure Model. Our main focus is about Distributed Registers, so our third contribution comes, we prove necessities and impossibilities in those models. In particular we prove that is it not possible to solve the weakest register specification in an asynchronous system. On the other side we prove lower bounds for the synchronous system, with respect to the proposed hierarchy models, and tight protocols to solve the Regular Register problem. To conclude, our last contribution is about the Approximate Agreement problem, a weaker form of Consensus. We solve such problem in the same round-based models as Consensus so far. The interesting result is the following, in presence of static Byzantine failures, lower bounds on the number of correct replicas does not change between consensus and approximate agreement. The same invariant still holds in presence of Mobile Byzantine failure. Moreover, along with lower bounds we propose a tight solution to solve approximate agreement.

京都大学 / 0048 / 新制・課程博士 / 博士(情報学) / 甲第19129号 / 情博第575号 / 新制||情||101(附属図書館) / 32080 / 京都大学大学院情報学研究科通信情報システム専攻 / (主査)教授 岩間 一雄, 教授 髙木 直史, 教授 五十嵐 淳 / 学位規則第4条第1項該当 / Doctor of Informatics / Kyoto University / DFAM

No description available.

<p dir="ltr">We analyze the query complexity of multiple search problems.</p><p dir="ltr">Firstly, we provide lower bounds on the complexity of "Local Search". In local search we are given a graph G and oracle access to a function f mapping the vertices to numbers, and seek a local minimum of f; i.e. a vertex v such that f(v) <= f(u) for all neighbors u of v. We provide separate lower bounds in terms of several graph parameters, including congestion, expansion, separation number, mixing time of a random walk, and spectral gap. To aid in showing these bounds, we design and use an improved relational adversary method for classical algorithms, building on the prior work of Scott Aaronson. We also obtain some quantum bounds using the traditional strong weighted adversary method.</p><p dir="ltr">Secondly, we show a multiplicative duality gap for Yao's minimax lemma by studying unordered search. We then go on to give tighter than asymptotic bounds for unordered and ordered search in rounds. Inspired by a connection through sorting with rank queries, we also provide tight asymptotic bounds for proportional cake cutting in rounds.</p>

Die vorliegende Arbeit besteht aus zwei Teilen. Der erste Teil der Arbeit befasst sich mit unteren Schranken für approximative Steinerbaumalgorithmen. Ein Steinerbaum ist ein kürzester Teilgraph, der eine gegebene Teilmenge der Knoten eines Graphen spannt. Das Berechnen eines Steinerbaumes ist ein klassisches NP-schweres Problem, und es existieren mehrere Approximationsalgorithmen, wobei bei den meisten Algorithmen die Approximationsgüte nur durch untere und obere Schranken eingegrenzt werden kann. Für einige dieser Algorithmen werden in dieser Arbeit Instanzen vorgestellt, welche die unteren Schranken verbessern. Für den Relativen Greedy Algorithmus wird außerdem ein Verfahren vorgestellt, mit dem die Güte des Algorithmus eingeschränkt auf die Graphenklasse mit k Terminalen auf einen beliebigen Faktor genau bestimmt werden kann. Der zweite Teil der Arbeit widmet sich vollständig bipartiten Subgraphen mit gleicher Partitionsgrößse, sogenannten balancierten Bicliquen. Seit langem ist bekannt, dass in dichten bipartiten Graphen balancierte Bicliquen mit Omega(log(n)) Knoten existieren, aber es ist unbekannt, ob und wie diese in polynomieller Zeit konstruiert werden können. Der zweite Teil liefert dazu einen Beitrag, indem ein polynomieller Algorithmus vorgestellt wird, der in hinreichend großen dichten bipartiten Graphen eine balancierte Biclique mit Omega(sqrt(log(n))) Knoten konstruiert. / This thesis consists of two parts. The first part is concerned with lower bounds for approximating Steiner trees. The Steiner tree problem is to find a shortest subgraph that spans a given set of vertices in a graph and is a classical NP-hard problem. Several approximation algorithms exist, but for most algorithms only lower and upper bounds for the approximation ratio are known. For some of these algorithms we present instances which improve the lower bounds. When the problem is restricted to the class of graphs with k terminals, we also present a method which can be used to determine the approximation ratio of the Relative Greedy Algorithm with arbitrary precision. The second part is about balanced bicliques, i.e. complete bipartite subgraphs with equal partition sizes. It has been known for a long time that every dense bipartite graph contains a balanced biclique of size Omega(log(n)), but whether and how such a biclique can be constructed in polynomial time is still unknown. Our contribution to this problem is a polynomial time algorithm which constructs a balanced biclique of size Omega(sqrt(log(n))) in sufficiently large and dense bipartite graphs.

Given a set of unit disks in the plane with union area A, what fraction of A can be covered by selecting a pairwise disjoint subset of the disks? Richard Rado conjectured 1/4 and proved 1/4.41. In this thesis, we consider a variant of this problem where the disjointness constraint is relaxed: selected disks must be k-colourable with disks of the same colour pairwise-disjoint. Rado's problem is then the case where k = 1, and we focus our investigations on what can be proven for k > 1. Motivated by the problem of channel-assignment for Wi-Fi wireless access points, in which the use of 3 or fewer channels is a standard practice, we show that for k = 3 we can cover at least 1/2.09 and for k = 2 we can cover at least 1/2.82. We present a randomized algorithm to select and colour a subset of n disks to achieve these bounds in O(n) expected time. To achieve the weaker bounds of 1/2.77 for k = 3 and 1/3.37 for k = 2 we present a deterministic O(n^2) time algorithm. We also look at what bounds can be proven for arbitrary k, presenting two different methods of deriving bounds for any given k and comparing their performance. One of our methods is an extension of the method used to prove bounds for k = 2 and k = 3 above, while the other method takes a novel approach. Rado's proof is constructive, and uses a regular lattice positioned over the given set of disks to guide disk selection. Our proofs are also constructive and extend this idea: we use a k-coloured regular lattice to guide both disk selection and colouring. The complexity of implementing many of the constructions used in our proofs is dominated by a lattice positioning step. As such, we discuss the algorithmic issues involved in positioning lattices as required by each of our proofs. In particular, we show that a required lattice positioning step used in the deterministic O(n^2) algorithm mentioned above is 3SUM-hard, providing evidence that this algorithm is optimal among algorithms employing such a lattice positioning approach. We also present evidence that a similar lattice positioning step used in the constructions for our better bounds for k = 2 and k = 3 may not have an efficient exact implementation.

Given a set of unit disks in the plane with union area A, what fraction of A can be covered by selecting a pairwise disjoint subset of the disks? Richard Rado conjectured 1/4 and proved 1/4.41. In this thesis, we consider a variant of this problem where the disjointness constraint is relaxed: selected disks must be k-colourable with disks of the same colour pairwise-disjoint. Rado's problem is then the case where k = 1, and we focus our investigations on what can be proven for k > 1. Motivated by the problem of channel-assignment for Wi-Fi wireless access points, in which the use of 3 or fewer channels is a standard practice, we show that for k = 3 we can cover at least 1/2.09 and for k = 2 we can cover at least 1/2.82. We present a randomized algorithm to select and colour a subset of n disks to achieve these bounds in O(n) expected time. To achieve the weaker bounds of 1/2.77 for k = 3 and 1/3.37 for k = 2 we present a deterministic O(n^2) time algorithm. We also look at what bounds can be proven for arbitrary k, presenting two different methods of deriving bounds for any given k and comparing their performance. One of our methods is an extension of the method used to prove bounds for k = 2 and k = 3 above, while the other method takes a novel approach. Rado's proof is constructive, and uses a regular lattice positioned over the given set of disks to guide disk selection. Our proofs are also constructive and extend this idea: we use a k-coloured regular lattice to guide both disk selection and colouring. The complexity of implementing many of the constructions used in our proofs is dominated by a lattice positioning step. As such, we discuss the algorithmic issues involved in positioning lattices as required by each of our proofs. In particular, we show that a required lattice positioning step used in the deterministic O(n^2) algorithm mentioned above is 3SUM-hard, providing evidence that this algorithm is optimal among algorithms employing such a lattice positioning approach. We also present evidence that a similar lattice positioning step used in the constructions for our better bounds for k = 2 and k = 3 may not have an efficient exact implementation.

In this thesis, a resource allocation problem in OFDMA is studied for the energy efficiency of wireless network. The objective is to minimize the total energy consumption which includes transmission energy consumption, and circuit energy consumption at both transmitter and receiver with required per user’s rate constraint. For problem solution, a heuristic algorithm with low computational complexity and suboptimal solution is proposed, developed in two steps with an increasing order of complexity. Besides, a bounding scheme based on model linearization of formulated nonlinear system model is also proposed to give lower and upper bounds for both small- and large-scale OFDMA network for further algorithm performance evaluation, while the implemented exhaustive search is only capable to provide the optimal solution for small-scale instance for algorithm performance evaluation. Numerical results show that the proposal heuristic algorithm can achieve near-optimal performance with applicable computational complexity even for large-scale networks, and that the bounds from the bounding scheme are very tight for both small- and large-scale OFDMA networks.

Modern applications of XML use automata operating on unranked trees. A common definition of tree automata operating on unranked trees uses a set of vertical states that define the bottom-up computation, and the transitions on vertical states are determined by so called horizontal languages recognized by finite automata on strings. The bottom-up computation of an unranked tree automaton may be either deterministic or nondeterministic, and further variants arise depending on whether the horizontal string languages defining the transitions are represented by DFAs or NFAs. There is also an alternative syntactic definition of determinism introduced by Cristau et al. It is known that a deterministic tree automaton with the smallest total number of states does not need to be unique nor have the smallest possible number of vertical states. We consider the question by how much we can reduce the total number of states by introducing additional vertical states. We give an upper bound for the state trade-off for deterministic tree automata where the horizontal languages are defined by DFAs, and a lower bound construction that, for variable sized alphabets, is close to the upper bound. We establish upper and lower bounds for the state complexity of conversions between different types of deterministic and nondeterministic unranked tree automata. The bounds are, usually, tight for the numbers of vertical states. Because a minimal deterministic unranked tree automaton need not be unique, establishing lower bounds for the number of horizontal states, that is, the combined size of DFAs used to define the horizontal languages, is challenging. Based on existing lower bound results for unambiguous finite automata we develop a lower bound criterion for the number of horizontal states. We consider the state complexity of operations on regular unranked tree languages. The concatenation of trees can be defined either as a sequential or a parallel operation. Furthermore, there are two essentially different ways to iterate sequential concatenation. We establish tight state complexity bounds for concatenation-like operations. In particular, for sequential concatenation and bottom-up iterated concatenation the bounds differ by an order of magnitude from the corresponding state complexity bounds for regular string languages. / Thesis (Ph.D, Computing) -- Queen's University, 2012-01-04 14:48:02.916

In this thesis, we show results for some well-studied problems from learning theory and combinatorial optimization. Learning Parities under the Uniform Distribution: We study the learnability of parities in the agnostic learning framework of Haussler and Kearns et al. We show that under the uniform distribution, agnostically learning parities reduces to learning parities with random classification noise, commonly referred to as the noisy parity problem. Together with the parity learning algorithm of Blum et al, this gives the first nontrivial algorithm for agnostic learning of parities. We use similar techniques to reduce learning of two other fundamental concept classes under the uniform distribution to learning of noisy parities. Namely, we show that learning of DNF expressions reduces to learning noisy parities of just logarithmic number of variables and learning of k-juntas reduces to learning noisy parities of k variables. Agnostic Learning of Halfspaces: We give an essentially optimal hardness result for agnostic learning of halfspaces over rationals. We show that for any constant ε finding a halfspace that agrees with an unknown function on 1/2+ε fraction of examples is NP-hard even when there exists a halfspace that agrees with the unknown function on 1-ε fraction of examples. This significantly improves on a number of previous hardness results for this problem. We extend the result to ε = 2[superscript-Ω(sqrt{log n})] assuming NP is not contained in DTIME(2[superscript(log n)O(1)]). Majorities of Halfspaces: We show that majorities of halfspaces are hard to PAC-learn using any representation, based on the cryptographic assumption underlying the Ajtai-Dwork cryptosystem. This also implies a hardness result for learning halfspaces with a high rate of adversarial noise even if the learning algorithm can output any efficiently computable hypothesis. Max-Clique, Chromatic Number and Min-3Lin-Deletion: We prove an improved hardness of approximation result for two problems, namely, the problem of finding the size of the largest clique in a graph (also referred to as the Max-Clique problem) and the problem of finding the chromatic number of a graph. We show that for any constant γ > 0, there is no polynomial time algorithm that approximates these problems within factor n/2[superscript(log n)3/4+γ] in an n vertex graph, assuming NP is not contained in BPTIME(2[superscript(log n)O(1)]). This improves the hardness factor of n/2[superscript (log n)1-γ'] for some small (unspecified) constant γ' > 0 shown by Khot. Our main idea is to show an improved hardness result for the Min-3Lin-Deletion problem. An instance of Min-3Lin-Deletion is a system of linear equations modulo 2, where each equation is over three variables. The objective is to find the minimum number of equations that need to be deleted so that the remaining system of equations has a satisfying assignment. We show a hardness factor of 2[superscript sqrt{log n}] for this problem, improving upon the hardness factor of (log n)[superscriptβ] shown by Hastad, for some small (unspecified) constant β > 0. The hardness results for Max-Clique and chromatic number are then obtained using the reduction from Min-3Lin-Deletion as given by Khot. Monotone Multilinear Boolean Circuits for Bipartite Perfect Matching: A monotone Boolean circuit is said to be multilinear if for any AND gate in the circuit, the minimal representation of the two input functions to the gate do not have any variable in common. We show that monotone multilinear Boolean circuits for computing bipartite perfect matching require exponential size. In fact we prove a stronger result by characterizing the structure of the smallest monotone multilinear Boolean circuits for the problem.