Introduction to Agent-Based Modeling
EES 4760/5760
Agent-Based and Individual-Based
Computational Modeling
Jonathan Gilligan
Class #1:
Monday, January 09
2024
Who Are You?
- Who are you? (Name, year, major)
- Computational skills (if any)
- programming, statistical analysis, …
- What do you want to get from this class?
- Ask me a question about computational modeling
- Something interesting about yourself
Textbooks
|
Steven F. Railsback & Volker Grimm, Agent-Based and
Individal-Based Modeling (2nd Edition)
|
|
Paul E. Smaldino, Modeling Social Behavior
|
Getting Started
For Monday:
- Download and install NetLogo on your computer.
- URL in syllabus and assignment sheet
Course Web Site
-
ees4760.jgilligan.org
- Syllabus
- All reading and homework assignments for the semester
- Slides from class.
- Files you will need for homework assignments.
- Links to helpful resources.
- Slides:
- The title slide has QR code with link to online version.
- PDF versions are also posted to course web site (link on title
slide)
- Slides have two-dimensional navigation (in a browser, hit “?” for
help)
Agent-Based Modeling
- Simulate individuals:
- Autonomous
- Heterogeneous
- Quasi-local
- Bounded rationality
- Simulate environment
- Emphasize simplicity, minimal assumptions
-
Emergence: Large-scale phenomena arise from
small-scale individual interactions
- Interesting when large-scale is not easily predicted from
small-scale
Simple Experiments
- Play with economics
Simple agents trade with each other
-
Confirm 1st welfare theorem:
Trading leads to Pareto equilibrium
-
Find conditions for satisfying theorem:
- Not necessary for traders to be completely rational
- How much rationality do you need?
- Equilibration can be slow
- Time-varying preferences can prevent equilibration
- Dynamics of agent-based models connect to nonlinear dynamics and
chaos
Economics of Cooperation
Game Theory
-
Prisoner’s Dilemma Game:
A Cooperates |
(3,3) |
(0,4) |
A Defects |
(4,0) |
(1,1) |
-
Nash Equilibrium:
- No matter what player A does, player B is better off defecting
- No matter what player B does, player A is better off defecting
-
End result: Both players end up worse off than if
they had both cooperated.
Iterated Prisoner’s Dilemma
- R. Axelrod (1981)
- Tournament of algorithms
- Winner: “tit-for-tat”
- Evolutionary Game Theory:
- Basic principles of good strategies:
- Be nice
- Be provocable
- Don’t be too envious
- Don’t be too clever
- Nay & Gilligan (2015)
- Real-world strategies involve randomness,
unpredictability
Example: Artificial
Anasazi
Axtell, Dean, Epstein, et
al.
![Long House Valley]()
Long House Valley (flourished ca. 1800 BCE–1300 CE)
Modeling Environment
![Artificial Anasazi Simulations]()
Constructing model
- Paleoclimate:
- Assess different kinds of soil
- Assess tree rings, pollen, etc.
- Reconstruct drought severity index
- Society:
- Archaeology gives #, location of households
- Make assumptions about:
- # people per household,
- Agriculture,
- …
- Devise rules for behavior:
- Marriage, reproduction, migration, …
- Simulate years 800–1300
Results
![Artificial Ansazi model results]()
Improvements
- Make agents heterogeneous
- Fit parameters to historical data
Results
![Artificial Ansazi improved model results]()
Introduction to Agent-Based Modeling
EES 4760/5760
Agent-Based and Individual-Based
Computational Modeling
Jonathan Gilligan
Class #1:
Monday, January 09
2024