Online Computation and Competitive Analysis

Similar documents
COMP Online Algorithms. Paging and k-server Problem. Shahin Kamali. Lecture 11 - Oct. 11, 2018 University of Manitoba

COMP Online Algorithms. Paging and k-server Problem. Shahin Kamali. Lecture 9 - Oct. 4, 2018 University of Manitoba

Online Call Control in Cellular Networks Revisited

Digital Signal Processing

Lecture Notes 3: Paging, K-Server and Metric Spaces

Lecture 7: The Principle of Deferred Decisions

THOMAS PANY SOFTWARE RECEIVERS

The tenure game. The tenure game. Winning strategies for the tenure game. Winning condition for the tenure game

Search then involves moving from state-to-state in the problem space to find a goal (or to terminate without finding a goal).

Game Theory and Randomized Algorithms

Joint Scheduling and Fast Cell Selection in OFDMA Wireless Networks

CONTENTS PREFACE. Part One THE DESIGN PROCESS: PROPERTIES, PARADIGMS AND THE EVOLUTIONARY STRUCTURE

Systems. Advanced Radar. Waveform Design and Diversity for. Fulvio Gini, Antonio De Maio and Lee Patton. Edited by

Cognitive Radio Techniques

Computational aspects of two-player zero-sum games Course notes for Computational Game Theory Section 3 Fall 2010

Contents. List of Figures List of Tables. Structure of the Book How to Use this Book Online Resources Acknowledgements

Benford's Law. Theory, the General Law of Relative Quantities, and Forensic Fraud Detection Applications. Alex Ely Kossovsky.

AI Approaches to Ultimate Tic-Tac-Toe

Hamming Codes as Error-Reducing Codes

Chapter 3 Learning in Two-Player Matrix Games

Analysis and Design of Autonomous Microwave Circuits

CS510 \ Lecture Ariel Stolerman

4.5. Latency in milliseconds Number of Shutdowns

Characteristics of Routes in a Road Traffic Assignment

each pair of constellation points. The binary symbol error that corresponds to an edge is its edge label. For a constellation with 2 n points, each bi

Department of Statistics and Operations Research Undergraduate Programmes

Systems Dependability Assessment

Wireless Network Coding with Local Network Views: Coded Layer Scheduling

Preface for Instructors and Other Teachers 1 About This Book... xvii

Introduction to Algorithms / Algorithms I Lecturer: Michael Dinitz Topic: Algorithms and Game Theory Date: 12/4/14

CS 1571 Introduction to AI Lecture 12. Adversarial search. CS 1571 Intro to AI. Announcements

Digital Communication Systems Engineering with

1.6 Congruence Modulo m

Stanford University CS261: Optimization Handout 9 Luca Trevisan February 1, 2011

@CRC Press. Discrete Mathematics. with Ducks. sarah-marie belcastro. let this be your watchword. serious mathematics treated with levity

Multivariate Permutation Tests: With Applications in Biostatistics

Algorithms. Abstract. We describe a simple construction of a family of permutations with a certain pseudo-random

3432 IEEE TRANSACTIONS ON INFORMATION THEORY, VOL. 53, NO. 10, OCTOBER 2007

Minimax Universal Sampling for Compound Multiband Channels

Adversarial Search. Rob Platt Northeastern University. Some images and slides are used from: AIMA CS188 UC Berkeley

18 Completeness and Compactness of First-Order Tableaux

Research Article n-digit Benford Converges to Benford

TOPOLOGY, LIMITS OF COMPLEX NUMBERS. Contents 1. Topology and limits of complex numbers 1

Novelty, Information and Surprise

SF2972: Game theory. Mark Voorneveld, February 2, 2015

Elements of Artificial Intelligence and Expert Systems

Adversarial Search. Human-aware Robotics. 2018/01/25 Chapter 5 in R&N 3rd Ø Announcement: Slides for this lecture are here:

Regents Physics Mr. Mellon Based on Chapter 22 and 23

Wireless Communications Over Rapidly Time-Varying Channels

LECTURE 26: GAME THEORY 1

Topic 1: defining games and strategies. SF2972: Game theory. Not allowed: Extensive form game: formal definition

MULTIRATE DIGITAL SIGNAL PROCESSING

Lower Bounds for the Number of Bends in Three-Dimensional Orthogonal Graph Drawings

Statistical Analysis of Nuel Tournaments Department of Statistics University of California, Berkeley

SIGNAL PROCESSING OF POWER QUALITY DISTURBANCES

Cooperative Wireless Charging Vehicle Scheduling

CS 188: Artificial Intelligence

3-2 Lecture 3: January Repeated Games A repeated game is a standard game which isplayed repeatedly. The utility of each player is the sum of

3644 IEEE TRANSACTIONS ON INFORMATION THEORY, VOL. 57, NO. 6, JUNE 2011

Permutations with short monotone subsequences

Near-Optimal Radio Use For Wireless Network Synch. Synchronization

Statistics, Probability and Noise

Broadcast in Radio Networks in the presence of Byzantine Adversaries

Online Frequency Assignment in Wireless Communication Networks

Signals, Sound, and Sensation

Hideo Okawara s Mixed Signal Lecture Series. DSP-Based Testing Fundamentals 6 Spectrum Analysis -- FFT

Preface... Acknowledgments... Glossary Introduction Management Systems and the Management of Chemical Process Safety...

Laboratory 1: Uncertainty Analysis

Alternation in the repeated Battle of the Sexes

Acentral problem in the design of wireless networks is how

Synchronization in Digital Communications

Precoding and Signal Shaping for Digital Transmission

Contents ACKNOWLEDGMENTS... LIST OF ACRONYMS... xvii PART 1. THINK UP A METHOD... 1

Practice Session 2. HW 1 Review

IEEE/ACM TRANSACTIONS ON NETWORKING, VOL. 17, NO. 6, DECEMBER /$ IEEE

CONTROL OF SENSORS FOR SEQUENTIAL DETECTION A STOCHASTIC APPROACH

Fundamentals of Industrial Control

Game Theory two-person, zero-sum games

Antonio Fernández Anta Dariusz R. Kowalski (U. of Liverpool) Miguel A. Mosteiro (Kean U. & U. Rey Juan Carlos) Prudence W. H. Wong (U.

Set 4: Game-Playing. ICS 271 Fall 2017 Kalev Kask

Automatic Control Motion control Advanced control techniques

CSCI 699: Topics in Learning and Game Theory Fall 2017 Lecture 3: Intro to Game Theory. Instructor: Shaddin Dughmi

Modular Arithmetic. claserken. July 2016

Lecture 2. 1 Nondeterministic Communication Complexity

Section Marks Agents / 8. Search / 10. Games / 13. Logic / 15. Total / 46

On the Benefit of Tunability in Reducing Electronic Port Counts in WDM/TDM Networks

Introduction to Electromagnetic Compatibility

Optimizing Media Access Strategy for Competing Cognitive Radio Networks Y. Gwon, S. Dastangoo, H. T. Kung

Formalising Event Reconstruction in Digital Investigations

How (Information Theoretically) Optimal Are Distributed Decisions?

Index Terms Deterministic channel model, Gaussian interference channel, successive decoding, sum-rate maximization.

Axiom A-1: To every angle there corresponds a unique, real number, 0 < < 180.

Bandwidth Estimation Using End-to- End Packet-Train Probing: Stochastic Foundation

Chapter 3 Chip Planning

1. Introduction to Game Theory

Universiteit Leiden Opleiding Informatica

Retransmission and Back-off Strategies for Broadcasting in Multi-hop Wireless Networks

Artificial Intelligence

Adversarial Search. Robert Platt Northeastern University. Some images and slides are used from: 1. CS188 UC Berkeley 2. RN, AIMA

AMORE meeting, 1-4 October, Leiden, Holland

Transcription:

Online Computation and Competitive Analysis Allan Borodin University of Toronto Ran El-Yaniv Technion - Israel Institute of Technology I CAMBRIDGE UNIVERSITY PRESS

Contents Preface page xiii 1 Introduction to Competitive Analysis: The List Accessing Problem 1 1.1 Basic Ideas and Terminology 1 1.2 The List Accessing Problem 4 1.3 The Sleator-Tarjan Result 6 1.4 The Potential Function Method 9 1.5 Some Lower Bounds 10 1.6 The List Factoring Technique 13 1.7 Historical Notes and Open Questions 19 2 Introduction to Randomized Algorithms: The List Accessing Problem 23 2.1 The Competitive Ratio of Randomized Algorithms 23 2.2 Algorithm BIT 24 2.3 Algorithm RMTF: Barely Random Versus Random 26 2.4 List Factoring-Phase Partitioning Revisited 27 2.5 COMB: An -Competitive Algorithm 29 2.6 Historical Notes and Open Questions 29 3 Paging: Deterministic Algorithms 32 3.1 Some Paging Algorithms 33 3.2 The (h, k)-paging Problem 34 3.3 List Accessing Algorithms as Paging Algorithms 35 3.4 LFD - An Optimal Offline Paging Algorithm 35 3.5 Marking and Conservative Algorithms and the Competitiveness of LRU, CLOCK, FIFO, and FWF 36 3.6 LIFO and LFU Are Not Competitive 39 3.7 The Full Access Cost Model 40 vii

viii CONTENTS 3.8 Theory Versus Practice 41 3.9 Historical Notes and Open Questions 42 4 Paging: Randomized Algorithms 44 4.1 Randomized Competitive Analysis 44 4.2 The Competitiveness of RANDOM 46 4.3 The MARK Algorithm 49 4.4 A Lower Bound for Randomized Paging Algorithms 51 4.5 Historical Notes and Open Questions 52 5 Alternative Paging Models: Beyond Pure Competitive Analysis 54 5.1 The Access Graph Model 54 5.2 Dynamic Access Graphs and Experimental Studies 65 5.3 Distributional Paging Models 68 5.4 Historical Notes and Open Questions 75 6 Game Theoretic Foundations 78 6.1 Games in Extensive and Strategic Forms 78 6.2 Randomized Strategies: Mixed, Behavioral, and General 83 6.3 Equivalence Theorems for Linear Games and Games of Perfect Recall 89 6.4 An Application to Paging and Competitive Analysis 93 6.5 Historical Notes and Open Questions 95 7 Request-Answer Games 98 7.1 Request-Answer Games 98 7.2 Randomized Adversaries 102 7.3 Relating the Adversaries 104 7.4 Historical Notes and Open Questions 107 8 Competitive Analysis and Zero-Sum Games 109 8.1 Two-Person Zero-Sum Games 109 8.2 On Generalizations of the Minimax Theorem for Infinite Games 114 8.3 Yao's Principle: A Technique for Obtaining Lower Bounds 115 8.4 Paging Revisited 120 8.5 Historical Notes 122 9 Metrical Task Systems 123 9.1 Formulation of (Metrical) Task Systems 123 9.2 An 8(iV - 1)-Competitive Traversal Algorithm 127 9.3 A2N-1 Lower Bound 128

CONTENTS IX 9.4 An Optimal Work Function MTS Algorithm 131 9.5 A Randomized Algorithm for a Uniform MTS 134 9.6 A Randomized Polylogarithmic Competitive Algorithm for Any MTS 135 9.7 Historical Notes and Open Questions 146 10 The ^-Server Problem 150 10.1 The Formulation of the Model 150 10.2 Some Basic Aspects of the ^-Server Problem 151 10.3 A Deterministic Lower Bound 153 10.4 ^-Servers on a Line and a Tree 155 10.5 An Efficient 3-Competitive 2-Server Algorithm for Euclidean Spaces 159 10.6 Balancing Algorithms 161 10.7 The ^-Server Work Function Algorithm 164 10.8 On Generalizations of the ^-Server Conjecture That Fail 175 10.9 Historical Notes and Open Questions 178 11 Randomized A>Server Algorithms 182 11.1 Oblivious Adversaries and Two Randomized ^-Server Algorithms for the Circle 182 11.2 A Lower Bound Against an Adaptive-Online Adversary 185 11.3 The-Cat and Rat Game and Applications to Randomized ^-Server Algorithms 186 11.4 The Harmonic Random Walk 191 11.5 The HARMONIC it-server Algorithm on an Arbitrary Metric Space 192 11.6 The Resistive Approach 196 11.7 Historical Notes and Open Questions 199 12 Load Balancing 201 12.1 Defining the Problem 201 12.2 Online Algorithms for Load Balancing of Permanent Jobs 204 12.3 Formulating the Machine Assignment Problem as a Generalized Virtual Circuit Routing Problem 210 12.4 Load Balancing of Temporary Jobs 213 12.5 Bin Packing 218 12.6 Historical Notes and Open Questions 222 13 Call Admission and Circuit Routing 226 13.1 Specifying the Problem 226

X CONTENTS 13.2 Throughput Maximization for Permanent Calls in Networks with Large Edge Capacities 227 13.3 Throughput Maximization for Limited Duration Calls 232 13.4 Experimental Results 234 13.5 Call Admission for Particular Networks: The Disjoint Paths Problem 237 13.6 The Disjoint Paths Problem: A Lower Bound for a Difficult Network 245 13.7 Routing on Optical Networks 250 13.8 Path Coloring for Particular Networks 253 13.9 A Lower Bound for Path Coloring on the Brick Wall Graph 259 13.10 Historical Notes and Open Problems 260 14 Search, Trading, and Portfolio Selection 264 14.1 Online Search and One-Way Trading 264 14.2 Online Portfolio Selection 273 14.3 Two-Way Trading: Statistical Adversaries and "Money Making" Algorithms 277 14.4 Two-Way Trading and the Fixed Fluctuation Model 281 14.5 Weighted Portfolio Selection Algorithms 290 14.6 Historical Notes and Open Questions 307 15 On Decision Theories and the Competitive Ratio 312 15.1 Certainty, Risk, and Strict Uncertainty 312 15.2 Decision Making Under Strict Uncertainty 315 15.3 The Competitive Ratio Axioms 321 15.4 Characterization of the Competitive Ratio 325 15.5 Characterizations of the Classical Criteria for Strict Uncertainty 333 15.6 An Example - The Leasing Problem 335 15.7 Decision Making Under Risk 339 15.8 Bayesian Approaches for Decision Making Under Uncertainty 346 15.9 Historical Notes and Open Questions 348 A Glossary 355 B Stochastic Analyses for List Accessing Algorithms 357 C The Harmonic Random Walk and Its Connection to Electrical Networks 361

CONTENTS XI D Proof of Lemmas 5.4 and 5.5 in Theorem 5.11: FAR Is a Uniformly Optimal Online Paging Algorithm 364 D.I Proof of Lemma 5.4: Type 1 Reps and the Construction of T' 364 D.2 Proof of Lemma 5.5: Type 2 Reps and the Construction of H 365 E Some Tools from Renewal Theory 369 E. 1 Renewal Processes 369 E.2 Wald's Equation 370 E.3 The Elementary Renewal Theorem 373 F Proof of Theorem 13.14: Disjoint Paths in an Array 375 F.I Short Distance Calls 375 F.2 Long Distance Calls 376 G Some Tools from the Theory of Types 379 H Two Technical Lemmas 382 Bibliography 389 Index 403