Artificial War: Multiagent-Based Simulation of Combat -- Contents -- Foreword -- Preface -- Acknowledgments -- Chapter 1 Introduction -- 1.1 Brief History of CNA's Complexity & Combat Research Project -- 1.1.1 The "Problem" -- 1.1.1.1 Nonlinear Dynamics -- 1.1.1.2 Complex Systems Theory -- 1.1.2 Applying the "New Sciences" to Warfare -- 1.1.2.1 Tier-I: General Metaphors for Complexity in War -- 1.1.2.2 Tier-II: Policy and General Guidelines for Strategy -- 1.1.2.3 Tier-III: Conventional Warfare Models and Approaches -- 1.1.2.4 Tier-IV: Description of the Complexity of Combat -- 1.1.2.5 Tier-V: Combat Technology Enhancement -- 1.1.2.6 Tier-VI: Combat Aids for the Battlefield -- 1.1.2.7 Tier-VII: Synthetic Combat Environments -- 1.1.2.8 Tier-VIII: Original Conceptualizations of Combat -- 1.1.2.9 Most Promising Applications -- 1.1.2.10 Why Land Warfare? -- 1.1.3 Warfare & Complexity -- 1.1.4 ISAAC -- 1.1.5 EINSTein -- 1.1.5.1 Portability Across Multiple Platforms -- 1.1.5.2 Availability -- 1.2 Background and Motivations -- 1.2.1 Lanchester Equations of Combat -- 1.2.1.1 Closed-Form Solutions -- 1.2.1.2 Limitations -- 1.2.2 Artificial Life -- 1.2.2.1 Boids -- 1.2.2.2 Decentralized Sorting -- 1.3 Models & Simulations: A Heuristic Discussion -- 1.3.1 Definitions -- 1.3.2 Connection to Reality -- 1.3.3 Mathematical Models -- 1.3.3.1 Minimal Idea Models -- 1.3.3.2 Minimum System Models -- 1.3.3.3 Synthetic System Models -- 1.3.4 Computer Simulations -- 1.3.5 What Price Complexity? -- 1.3.5.1 Tierra -- 1.4 Combat Simulation -- 1.4.1 Modeling and Simulation Master Plan -- 1.4.2 Modeling Human Behavior and Command Decision-Making -- 1.4.3 Conventional Simulations -- 1.4.3.1 ELAN -- 1.4.3.2 JANUS -- 1.4.3.3 CASTFOREM -- 1.4.3.4 ModSAF -- 1.4.4 Future of Modeling Technology -- 1.5 Multiagent-Based Models and Simulations -- 1.5.1 Autonomous Agents.

1.5.2 How is Multiagent-Based Modeling Really Done? -- 1.5.3 Agent-Based Simulations vs. Traditional Mathematical Models -- 1.5.3.1 Synthesist Approach -- 1.5.3.2 Explanatory Tools -- 1.5.4 Multiagent-Based Simulations vs. Traditional AI -- 1.5.5 Examples of MultiAgent-Based Simulations -- 1.5.5.1 SWARM -- 1.5.5.2 General Purpose Simulations -- 1.5.6 Value of Multiagent-Based Simulations -- 1.5.7 CA-Based & Other EINSTein-Related Combat Models -- 1.5.7.1 CROCADILE -- 1.5.7.2 DEXES -- 1.5.7.3 MANA -- 1.5.7.4 Minimalist Models -- 1.5.7.5 SEM -- 1.5.7.6 Socrates -- 1.5.7.7 SWarrior -- 1.5.7.8 THOR -- 1.6 EINSTein as an Exemplar of More General Models of Complex Adaptive Systems -- 1.6.1 Persian Gulf Scenario -- 1.6.2 SCUDHunt -- 1.6.3 Social Modeling: Riots and Civil Unrest -- 1.6.4 General Applications -- 1.6.5 Universal Patterns of Behavior -- 1.7 Goals & Payoffs for Developing EINSTein -- 1.7.1 Command & Control -- 1.7.2 Pattern Recognition -- 1.7.3 "What If ?" Experimentation -- 1.7.4 Fundamental Grammar of Combat? -- 1.8 Toward an Axiological Ontology of Complex Systems -- 1.8.1 Why "Value"? -- 1.8.2 Why "Axiological Ontology"? -- Chapter 2 Nonlinear Dynamics, Deterministic Chaos and Complex Adaptive Systems: A Primer -- 2.1 Nonlinear Dynamics and Chaos -- 2.1.1 Brief History -- 2.1.2 Dynamical Systems -- 2.1.2.1 Poincare Maps -- 2.1.2.2 Phase Space Volumes -- 2.1.2.3 Dissipative Dynamical Systems (A < 0) -- 2.1.2.4 Conservative Systems (A < 0 ) -- 2.1.3 Deterministic Chaos -- 2.1.3.1 Example #1: The Bernoulli Shift Map -- 2.1.3.2 Example #2: The Logistic Map -- 2.1.4 Qualitative Characterization of Chaos -- 2.1.4.1 Time Dependence -- 2.1.4.2 Poincare Maps -- 2.1.4.3 Autocorrelation Function -- 2.1.4.4 Power Spectrum -- 2.1.5 Quantitative Characterization of Chaos -- 2.1.5.1 Lyapunov Exponents -- 2.1.5.2 Entropies and Dimensions.

2.1.6 Time-Series Forecasting and Predictability -- 2.1.7 Chaotic Control -- 2.2 Complex Adaptive Systems -- 2.2.1 References -- 2.2.2 Short History -- 2.2.3 General Properties: A Heuristic Discussion -- 2.2.4 Measures of Complexity -- 2.2.4.1 Complexity as information -- 2.2.4.2 Complexity of a Graph -- 2.2.4.3 Complexity of a Simplex -- 2.2.4.4 Complexity of Hierarchical Systems -- 2.2.4.5 Computational Complexity -- 2.2.4.6 Algorithmic Complexity -- 2.2.4.7 Logical Depth -- 2.2.4.8 Thermodynamic Depth -- 2.2.5 Complexity as Science: Toward a New Worldview? -- 2.2.6 Artificial Life -- 2.2.6.1 Self-Reproducing Automata -- 2.2.6.2 General Properties -- 2.2.7 Cellular Automata -- 2.2.7.1 One-dimensional CA -- 2.2.7.2 Two-dimensional CA -- 2.2.7.3 Other Kinds of CA -- 2.2.8 Self- Organized Criticality -- 2.2.8.1 Properties -- 2.2.8.2 Sandpiles -- 2.2.8.3 Example: One-dimensional CA Sandpile -- Chapter 3 Nonlinearity, Complexity, and Warfare: Eight Tiers of Applicability -- 3.1 Approach -- 3.2 Tier I: General Metaphors for Complexity in War -- 3.2.1 What is a Metaphor? -- 3.2.2 Metaphors and War -- 3.2.3 Metaphor Shift -- 3.2.3.1 Nonlinearity -- 3.2.3.2 Deterministic Chaos -- 3.2.3.3 Complexity -- 3.2.3.4 Self-Organization -- 3.2.3.5 Emergence -- 3.3 Tier II: Policy and General Guidelines for Strategy -- 3.3.1 What Does the New Metaphor Give Us? -- 3.3.2 Policy -- 3.3.3 Organizational Structure -- 3.3.4 Intelligence Analysis -- 3.3.5 Policy Exploitation of Characteristic Time Scales of Combat -- 3.4 Tier III: "Conventional" Warfare Models and Approaches -- 3.4.1 Testing for the Veracity of Conventional Models -- 3.4.2 Non-Monoticities and Chaos -- 3.4.3 Minimalist Modeling -- 3.4.4 Generalizations of Lanchester 's equations -- 3.4.4.1 Adaptive Dynamic Model of Combat -- 3.4.4.2 Lotka- Volterra Equations.

3.4.5 Nonlinear Dynamics and Chaos in Arms-Race Models -- 3.5 Tier IV: Description of the Complexity of Combat -- 3.5.1 Attractor Reconstruction from Time-Series Data -- 3.5.2 Fractals and Combat -- 3.5.3 Evidence of Chaos in War From Historical Data? -- 3.5.4 Evidence of Self- Organized Criticality From Historical Data? -- 3.5.4.1 Combat Casualties -- 3.5.4.2 Message Traffic -- 3.5.5 Use of Complex Systems Inspired Measures to Describe Combat -- 3.5.5.1 Casualty-Based Entropy -- 3.5.6 Use of Relativistic Information to Describe Command and Control Processes -- 3.6 Tier V: Combat Technology Enhancement -- 3.6.1 Computer Viruses ("computer counter-measures ") -- 3.6.2 Fractal Image Compression -- 3.6.3 Cryptography -- 3.7 Tier VI: Combat Aids -- 3.7.1 Using Genetic Algorithms to Evolve Tank Strategies -- 3.7.1.1 Smart-Tank's Chromosome -- 3.7.2 Tactical Decision Aids -- 3.7.3 Classifier Systems -- 3.7.4 How can Genetic Algorithms be Used? -- 3.7.5 Tactical Picture Agents -- 3.8 Tier VII: Synthetic Combat Environments -- 3.8.1 Combat Simulation using Cellular Automata -- 3.8.2 Multiagent-Based Simulations -- 3.9 Tier VIII: Original Conceptualizations of Combat -- 3.9.1 Dueling Parasites -- 3.9.2 Percolation Theory and Command and Control Processes -- 3.9.3 Exploiting Chaos -- 3.9.3.1 Chaotic Control -- 3.9.3.2 Chaotic Synchronization -- 3.9.3.3 Taming Chaos -- 3.9.4 Pattern Recognition -- 3.9.4.1 High-Level Rule Extraction -- 3.9.4.2 Self-0rganizing Maps -- 3.9.4.3 Data-Base Mining for Knowledge -- 3.9.5 Fire-Ant Warfare -- Chapter 4 EINSTein: Mathematical Overview -- 4.1 Introduction -- 4.2 Design Philosophy -- 4.2.1 Agent Hierarchy -- 4.2.1.1 Combat Agent Level -- 4.2.1.2 Local Command Level -- 4.2.1.3 Global Command Level -- 4.2.1.4 Supreme Command Level -- 4.2.2 Guiding Principles -- 4.2.2.1 Simplicity -- 4.2.2.2 Consistency.

4.3 Abstract Agent Architecture -- 4.3.1 Overview -- 4.3.2 Dynamics of Value -- 4.3.3 General Formalism -- 4.3.3.1 Environment -- 4.3.3.2 Agents -- 4.3.3.3 Relations -- 4.3.3.4 Actions -- 4.3.3.5 Action Selection Rules -- 4.3.4 Agents in EINSTein -- 4.3.5 Actions -- 4.3.5.1 Action Selection -- 4.3.5.2 Personality -- 4.3.5.3 Motivations -- 4.3.6 Features -- 4.3.7 Local Context -- 4.3.8 Example -- 4.3.9 Ontological Partitioning -- 4.3.10 Communication -- 4.3.11 Axiological Ontology -- 4.3.12 Preventing a Combinatorial Explosion -- 4.3.12.1 Finite State Machines -- 4.3.12.2 Behavior-Based Action Selection -- 4.3.12.3 Layered Architecture -- Color Plates -- Chapter 5 EINSTein: Methodology -- 5.1 Program Structure -- 5.1.1 Source Code -- 5.1.2 Object-Oriented -- 5.1.2.1 Objects -- 5.1.2.2 Classes -- 5.1.3 Program Flow -- 5.2 Combat Engine -- 5.2.1 Agents -- 5.2.2 Battlefield -- 5.2.3 Agent Sensor Parameters -- 5.2.3.1 Attention Range -- 5.2.3.2 Combat Range -- 5.2.3.3 Communications Range -- 5.2.3.4 Fire Range -- 5.2.3.5 Movement Range -- 5.2.3.6 Sensor Range -- 5.2.4 Agent Personalities -- 5.2.5 Agent Action Selection -- 5.2.5.1 General Logic -- 5.2.5.2 Environmental Context -- 5.2.5.3 Engineering Agents vs. Engineering Massions -- 5.2.5.4 Fundamental Axioms of EINSTein's Action Selection Logic -- 5.2.5.5 Penalty Function -- 5.2.5.6 Battlefield Boundary Weight -- 5.2.5.7 Area Weight -- 5.2.5.8 Formation Weight -- 5.2.5.9 Inter-Squad Connectivity Weight -- 5.2.5.10 Intra-Squad Connectivity Weight -- 5.2.5.11 Local Command Weight -- 5.2.5.12 Local Command Obey Weight -- 5.2.5.13 Terrain Weight -- 5.2.5.14 Example -- 5.2.6 Move Decision Logic Flags -- 5.2.6.1 Sampling Order -- 5.2.6.2 Curtailment Flags -- 5.2.7 Meta-Rules -- 5.2.7.1 Advance -- 5.2.7.2 Cluster -- 5.2.7.3 Combat -- 5.2.7.4 Hold Position -- 5.2.7.5 Pursuit-I (Pursuit Off).

5.2.7.6 Pursuit-II (Exclusive Pursuit On).