The principles of agent-based modeling

Document Sample
The principles of agent-based modeling Powered By Docstoc
					Advanced Computational Modeling
of Social Systems
           Lars-Erik Cederman and Luc Girardin
  Center for Comparative and International Studies (CIS)
   Swiss Federal Institute of Technology Zurich (ETH)
           Today‘s agenda

• Course goals
• Introduction to ABM
• Course logistics
             Course goals

•   Study the principles of agent-based modeling
•   Survey applications to the social sciences
•   Develop your own computational model of a social system
•   Prerequisite: Programming skills
             Four types of models

                             Modeling language:
                         Deductive      Computational

                         1. Analytical           2. Macro-
                         macro models            simulation

           Micro-         3. Rational          4. Agent-based
       mechanisms            choice               modeling
             1. Analytical macro models

• Equilibrium conditions or systemic
  variables traced in time
• Closed-form, and often based on
  differential equations
• Examples: macro economics and
  traditional systems theory
               2. Macro simulation

• Dynamic systems, tracing macro variables
  over time
• Based on simulation
• Systems theory and Global Modeling

    Jay Forrester, MIT
             3. Rational choice modeling

• Individualist reaction to macro
• Decision theory and game theory
• Analytical equilibrium solutions
• Used in micro-economics and spreading
  to other social sciences
              4. Agent-based modeling

• ABM is a computational methodology that allows the analyst to create,
  analyze, and experiment with, artificial worlds populated by agents that
  interact in non-trivial ways
• Bottom-up
• Computational
• Builds on CAs and DAI
    Disaggregated modeling

 else                        Inanimate agents
            Animate agents                      Observer


                     Organizations of agents
                    Artificial world
            Microeconomics  ABM

Analytical             Synthetic approach
Equilibrium            Non-equilibrium theory
Nomothetic             Generative method
Variable-based         Configurative ontology
                Analytical 
                                   Synthetic approach    11

• Hope to solve problems through strategy of “divide
  and conquer”
• Need to make ceteris paribus assumption
• But in complex systems this assumption breaks down
• Herbert Simon: Complex systems are composed of
  large numbers of parts that interact in a non-linear
• Need to study interactions explicitly
                Equilibrium 
                             Non-equilibrium theory          12

• Standard assumption in the social sciences:
  “efficient” history
• But contingency and positive feedback undermine
  this perspective
• Complexity theory and non-equilibrium physics
• Statistical regularities at the macro level despite
  micro-level contingency

                          Example: Avalanches in rice pile
                Nomothetic         
                                   Generative method   13

• Search for causal regularities
• Hempel’s “covering laws”
• But what to do with complex social systems that
  have few counterparts?
• Scientific realists explain complex patterns by
  deriving the mechanisms that generate them
• Axelrod: “third way of doing science”
• Epstein: “if you can’t grow it, you haven’t
  explained it!”
                    Variable-based 
                                   Configurative ontology   14

• Conventional models are variable-based
• Social entities are assumed implicitly
• But variables say little about social forms
• A social form is a configuration of social interactions
  and actors together with the structures in which they
  are embedded
• ABM good at endogenizing interactions and actors
• Object-orientation is well suited to capture agents
            A third way of doing science

1. Deduction
     – Derive theorems from assumptions
2. Induction
     – Find patterns in empirical data
3. Simulation
     – Start with explicit assumptions (deduction)
     – Generate data suitable for analysis (induction)
            Empirical understanding

• Why have particular large-scale regularities evolved and persisted,
  even when there is little top-down control?
• Examples: standing ovations, trade networks, socially accepted
  monies, mutual cooperation based on reciprocity, and social
• ABM: seek causal explanations grounded in the repeated
  interactions of agents operating in specified environments
             Normative understanding

• How can agent-based models be used as laboratories for the
  discovery of good designs?
• Examples: design of auction systems, voting rules, and law
• ABM: evaluate whether designs proposed for social policies,
  institutions, or processes will result in socially desirable system
  performance over time

• How can greater insight be attained about the fundamental
  causal mechanisms in social systems?
• Examples: city segregation (or “tipping”) model developed by
  Thomas Schelling
• The large-scale effects of interacting agents are often surprising
  because it can be hard to anticipate the full consequences of even
  simple forms of interaction
            Methodological advancement

• How to provide ABM researchers with the methods and tools
  they need of social systems through controlled computational
• Examples: methodological principles, programming tools,
  visualization techniques
             A methodological approach

• ABM is a methodological approach that could ultimately permit
  two important developments:
   – The rigorous testing, refinement, and extension of existing theories
     that have proved to be difficult to formulate and evaluate using
     standard statistical and mathematical tools
   – A deeper understanding of fundamental causal mechanisms in multi-
     agent systems whose study is currently separated by artificial
     disciplinary boundaries

• Performance evaluation
   – Class participation
   – Class presentation
   – Term paper
• Readings
   – On our server
• Class home page:
                       Course schedule

     – 29.03.2005: Introduction and logistics
•   Concepts
     – 05.04.2005: Complexity theory
     – 12.04.2005: Artificial life and intelligence
     – 19.04.2005: Network models
•   Applications
     –    26.04.2005: Traffic   Project memo due!
     –    03.05.2005: Economy
     –    10.05.2005: Sociology
     –    17.05.2005: Conflict
•   Empirical methods
     – 24.05.2005: Validation
     – 31.05.2005: GIS
•   Student presentations
     – 07.06; 14.06; 21.06; 28.06.2005
•   Final paper due July 5, 2005
                 Complexity theory

Complex adaptive systems exhibit properties
that emerge from local interactions among
many heterogeneous agents mutually
constituting their own environment

                                                         A model of the Internet
                                The Santa Fe Institute
               Complex Adaptive Systems

A CAS is a network exhibiting aggregate properties that emerge from
primarily local interaction among many, typically heterogeneous agents
mutually constituting their own environment.

       Emergent properties
       Large numbers of diverse agents
       Local and/or selective interaction
       Adaptation through selection
       Endogenous, non-parametric environment

Shared By: