Agent-Based Modelling of Socio-Technical Systems (eBook)

Agent-Based Modelling of Socio-Technical Systems 4
Foreword 6
Preface 9
Contents 11
Contributors 17
List of Figures 23
List of Tables 26
Chapter 1: Introduction 27
1.1 Why This Book? 27
1.2 Infrastructures as Complex Adaptive Socio-technical Systems 28
1.3 Better Decision-Making Needed 30
1.4 Agent-Based Modelling for Decision Support 31
1.5 A Book in Two Parts 33
Acknowledgements 33
References 33
Part I: Theory and Practice 35
Chapter 2: Theory 36
2.1 Introduction 36
2.1.1 Focus 37
2.1.2 Structure of the Chapter 37
2.1.3 Example: Westland Greenhouse Cluster 38
2.2 Systems 39
2.2.1 History of Systems Thinking 39
Greenhouse Example 40
2.2.2 Systems 41
Idealisation 42
Multiple Components 42
Components Are Interdependent 42
Organised 43
Emergent Properties 43
Boundaries 43
Enduring 43
Environment 44
Feedback 44
Non-trivial Behaviour 44
2.2.3 World Views 44
Greenhouse Example 45
2.2.4 Observer-Dependence 46
Objectivity 46
Greenhouse Example 47
Reductionism and Holism 48
Greenhouse Example 49
2.2.5 System Boundaries 49
2.2.6 System Nestedness 50
Greenhouse Example 51
2.3 Adaptive 51
2.3.1 Adaptation Versus Evolution 52
2.3.2 Evolution-More than just Biology 53
2.3.3 Adaptation in Its Many Forms 55
2.3.4 Direction of Adaptation 56
2.3.5 Coupled Fitness Landscape 57
Irreversibility 59
2.3.6 Intractability 59
2.4 Complexity 61
2.4.1 Simple 61
Functional Simplicity 62
Structural Simplicity 63
Occam's Razor 63
Greenhouse Example 64
2.4.2 Complicated 64
Greenhouse Example 66
2.4.3 Complex 66
Dynamics 67
Self-similarity or Scale Invariance 68
Greenhouse Example 69
2.5 Complex Adaptive Systems 69
Greenhouse Example 70
2.5.1 Chaos and Randomness 70
Repetition 71
Deterministic 71
Initial Conditions 71
Attractors 72
Instability and Robustness 72
Greenhouse Example 73
2.5.2 Emergence, Self-organisation and Patterns 73
Greenhouse Example 74
Self-organisation 75
Patterns 75
2.6 Modelling Complex Adaptive Systems 76
2.6.1 What Does a Model of a Complex Adaptive System Need? 76
Multi-domain and Multi-disciplinary Knowledge 77
Generative and Bottom up Capacity 78
Adaptivity 78
Modelling Options 78
2.6.2 Agent-Based Modelling 79
2.6.3 What It Is and Is not 80
Agent-Based Model 80
Multi-agent System 81
Arti?cial Intelligence 81
Object-Oriented Program? 82
2.7 Anatomy of an Agent-Based Model 82
2.7.1 Agent 82
2.7.1.1 State 83
2.7.1.2 Changing States 84
Rules 84
Actions 85
Behaviour 85
Greenhouse Example 85
2.7.2 Environment 86
2.7.2.1 Information 86
2.7.2.2 Structure 87
Soup 88
Space 88
Small-World Networks 89
Scale-Free Networks 89
Greenhouse Example 90
2.7.3 Time 90
Discrete Time 91
Assumption of Parallelism 91
Scheduler 91
Greenhouse Example 92
References 93
Chapter 3: Practice 97
3.1 Introduction 97
3.2 Step 1: Problem Formulation and Actor Identi?cation 98
3.2.1 Step 1 Example 99
Example: Westland Greenhouse Cluster 100
What Is the Problem 100
Initial Hypothesis 100
Whose Problem Are We Addressing? 100
Other Actors 100
Our Role 101
3.3 Step 2: System Identi?cation and Decomposition 101
3.3.1 Inventory 101
3.3.2 Structuring 102
3.3.2.1 Structuring of Agents and Interactions 103
3.3.2.2 Iteration 103
3.3.2.3 Environment 104
3.3.3 Step 2 Example 105
3.4 Step 3: Concept Formalisation 106
3.4.1 Software Data Structures 107
3.4.2 Ontology 108
3.4.3 Step 3 Example 109
Software Data Structures 109
Ontology 110
3.5 Step 4: Model Formalisation 112
3.5.1 Developing a Model Narrative 112
3.5.2 Pseudo-code 113
3.5.2.1 Elements of Pseudo-code 113
Computation and Assignment 113
Iterations and Loops 114
Conditions 114
Input/Output 115
3.5.2.2 Uni?ed Modelling Language 115
3.5.3 Step 4 Example 116
Model Narrative Example, Actions per Time Tick 116
Pseudo-code 116
3.6 Step 5: Software Implementation 117
3.6.1 Modelling Environment 118
3.6.1.1 NetLogo 118
3.6.1.2 Repast 118
3.6.1.3 Custom Code 119
3.6.2 Programming Practices 119
3.6.2.1 Version Control 120
3.6.2.2 Documenting Code 120
3.6.2.3 Naming Conventions 121
3.6.2.4 Divisions of Tasks and Responsibilities 121
3.6.2.5 Bug Tracking 121
3.6.3 Step 5 Example 122
NetLogo 122
3.7 Step 6: Model Veri?cation 122
3.7.1 Recording and Tracking Agent Behaviour 124
3.7.2 Single-Agent Testing 125
3.7.2.1 Theoretical Prediction and Sanity Checks 125
3.7.2.2 Breaking the Agent 126
3.7.3 Interaction Testing in a Minimal Model 126
3.7.4 Multi-agent Testing 127
3.7.4.1 Variability Testing 127
3.7.4.2 Timeline Sanity 128
3.7.5 Step 6 Example 128
3.8 Step 7: Experimentation 129
3.8.1 Experiment Design Aspects 129
3.8.1.1 Hypothesis Type 129
3.8.1.2 Time 131
3.8.1.3 Scenarios and Scenario Space 131
3.8.2 Experiment Setup 132
3.8.2.1 Full Factorial 132
3.8.2.2 Random Parameter 133
3.8.2.3 Latin Hypercube Sampling 133
3.8.2.4 Monte Carlo 134
3.8.2.5 Repetitions 134
3.8.2.6 Random Seed 135
3.8.3 Experiment Execution 136
3.8.3.1 Running on a Single Computer 136
3.8.3.2 Scaling, Running on Clusters 137
3.8.3.3 Collecting and Storing Data 137
3.8.3.4 Checking Data Consistency 138
3.8.4 Step 7 Example 138
Hypothesis Example 138
3.9 Step 8: Data Analysis 140
3.9.1 Data Exploration 140
3.9.1.1 Hypothesis Driven Analysis 141
3.9.1.2 Location of Patterns 141
3.9.1.3 Data Analysis 142
3.9.1.4 Analysis Tools 142
3.9.2 Pattern Visualisation and Identi?cation 144
3.9.2.1 Recognising Emergent Patterns 144
Dynamic Behaviour 144
Attractor Changes 144
Metastable Behaviour 144
Lack of a Pattern 144
3.9.2.2 Visualisation 144
Network Representation 146
Animation 146
3.9.2.3 Visualisation Caveats 146
3.9.3 Pattern Interpretation and Explanation 147
3.9.4 Experiment Iteration 147
3.9.5 Step 8 Example 148
Hypothesis Driven Analysis 148
3.10 Step 9: Model Validation 150
3.10.1 Historic Replay 151
3.10.2 Expert Validation 152
3.10.3 Validation by Literature Comparison 153
3.10.4 Validation by Model Replication 153
3.10.5 Step 9 Example 154
Expert Validation 154
3.11 Step 10: Model Use 154
3.11.1 Outcome Presentation 155
3.11.2 Raising New Questions 155
3.11.3 Long Term Stakeholder Engagement 156
3.11.4 Agent-Based Models and Stakeholders 156
3.11.5 Computer Models and Mental Models 157
3.11.6 Step 10 Example 158
3.12 Chapter Conclusions 159
References 160
Part II: Case Studies 162
Chapter 4: Introduction to the Case Studies 163
4.1 Case Studies 163
4.1.1 Operational Models 164
4.1.2 Tactical Models 164
4.1.3 Strategic Models 164
4.1.4 Setup of Case Study Chapters 165
4.1.5 State-of-the-Art Modelling 166
4.2 An Ontology for Socio-technical Systems 166
4.2.1 Aims 166
4.2.2 Development 167
4.2.3 Key Concepts 168
4.2.4 Usage 169
References 171
Chapter 5: Agent-Based Models of Supply Chains 172
5.1 Introduction 172
5.1.1 Supply Chains 173
5.1.2 Abnormal Situation Management 173
5.1.3 Supply Chain Modelling 174
5.1.4 Chapter Organisation 175
5.2 Oil Re?nery Supply Chain 175
5.2.1 Step 1: Problem Formulation and Actor Identi?cation 175
5.2.2 Step 2: System Identi?cation and Decomposition 176
5.3 Multi-plant Enterprise Supply Chain 180
5.3.1 Step 1: Problem Formulation and Actor Identi?cation 180
5.3.2 Step 2: System Identi?cation and Decomposition 181
5.4 Step 3: Concept Formalisation 181
5.5 Step 4: Model Formalisation 185
5.5.1 Procurement 185
5.5.2 Order Assignment 186
5.6 Step 5: Software Implementation 188
5.7 Step 6: Model Veri?cation 188
5.8 Step 7: Experimentation 190
5.8.1 Experimental Setup for the Oil Re?nery Supply Chain 190
5.8.2 Experimental Setup for the Multi-plant Enterprise 193
5.9 Step 8: Data Analysis 194
5.9.1 Delay in Shipment in the Oil Re?nery Supply Chain 194
5.9.2 Normal and Abnormal Behaviour Analysis for the Multi-plant Enterprise 195
5.10 Step 9: Model Validation 198
5.11 Step 10: Model Use 198
5.12 Conclusions 199
References 200
Chapter 6: An Agent-Based Model of Consumer Lighting 202
6.1 Introduction 202
6.2 Step 1: Problem Formulation and Actor Identi?cation 204
6.3 Step 2: System Identi?cation and Decomposition 204
6.3.1 Inventory 204
6.3.1.1 Social Subsystem 205
6.3.1.2 Technological Subsystem 205
6.3.1.3 Interactions 206
Social Interactions 206
Technological Interactions 207
Socio-technical Interactions 208
6.3.2 Structuring 208
6.4 Step 3: Concept Formalisation 209
6.5 Step 4: Model Formalisation 211
6.5.1 Social Structure 211
6.5.2 Model Narrative and Pseudo Code 212
6.6 Step 5: Software Implementation 214
6.7 Step 6: Model Veri?cation 215
6.8 Step 7: Experimentation 215
6.9 Step 8: Data Analysis 216
6.10 Step 9: Model Validation 218
6.11 Step 10: Model Use 219
6.12 Conclusions 219
References 220
Chapter 7: An Agent-Based Model of CO2 Policies and Electricity Generation 222
7.1 Introduction 222
7.2 Step 1: Problem Formulation and Actor Identi?cation 223
7.3 Step 2: System Identi?cation and Decomposition 225
7.4 Step 3: Concept Formalisation 227
7.5 Step 4: Model Formalisation 227
7.5.1 Model Narrative 228
7.5.2 Investment 229
7.5.3 Bidding on Markets 229
7.6 Step 5: Software Implementation 231
7.7 Step 6: Model Veri?cation 232
7.7.1 Preliminary Simulations 232
7.7.2 Extreme-Condition Tests and Discussion 233
7.7.3 Agent Behaviour Tests 233
7.7.4 Repetitions 234
7.8 Step 7: Experimentation 234
7.9 Step 8: Data Analysis 235
7.10 Step 9: Model Validation 236
7.11 Step 10: Model Use 236
7.11.1 Emergent Insights from Iterations and Discussions 237
7.11.2 Serious Game 237
7.12 Conclusions 238
References 239
Chapter 8: An Agent-Based Model of a Mobile Phone Production, Consumption and Recycling Network 241
8.1 Introduction 241
8.2 Step 1: Problem Formulation and Actor Identi?cation 243
8.3 Step 2: System Identi?cation and Decomposition 243
8.4 Step 3: Concept Formalisation 246
8.5 Step 4: Model Formalisation 248
8.5.1 Create Contracts 248
8.5.2 Purchase 249
8.5.3 Process 249
8.5.4 Invest 251
8.6 Step 5: Software Implementation 251
8.7 Step 6: Model Veri?cation 251
8.8 Step 7: Experimentation 253
8.9 Step 8: Data Analysis 255
8.9.1 Default Case 255
8.9.2 Sweep 1 256
8.9.3 Sweep 2 257
8.9.4 Sweep 3 259
8.10 Step 9: Model Validation 259
8.11 Step 10: Model Use 260
8.12 Conclusions 261
References 262
Chapter 9: Next Steps in Modelling Socio-technical Systems: Towards Collaborative Modelling 264
9.1 Complications of Modelling Socio-technical Systems 264
9.1.1 Agent-Based Model as Software 265
9.2 Applying Semantic Web Technologies and Methods for Modelling 266
9.2.1 Semantic Web Technologies 266
9.2.2 Using SPARQL to Extract Structured Data 268
9.2.3 Using SPARQL Within Simulations 269
9.2.3.1 SPARQL for Agent Intelligence 270
9.2.3.2 SPARQL for Simulation Debugging 271
9.2.3.3 Disadvantages 272
9.2.4 Implementations by Other Researchers 272
9.3 Case Study: Mobile Phone Recycling Networks 272
9.4 Future Work: Enabling Collaboration Between Modellers 276
9.4.1 Limitations of Using a Single Ontology for Modelling 276
9.4.2 Enabling Multiple System Views 277
9.4.3 Integrating Multiple Ontologies Together for Simulations 278
9.4.4 Simple Example of Integrating Multiple Ontologies 278
9.5 Conclusion 281
References 281
Index 283