CS426

Download as PDF

Agent-based Modeling and Simulation

Course (UG/PG)

Undergraduate

Offering Unit/Department

Course Description

In this course, we introduce agent-based modeling and simulation (ABMS) as an approach for studying complex business and social processes. With examples from domains such as transportation, economics, finance, and urban planning, we show how ABMS can help us better understand complex business and social phenomena. ABMS systems are particularly powerful if we want to describe a system populated by many independent and heterogeneous decision makers (who can be collaborators or competitors). ABMS systems can also be used in performing policy evaluations and generating decision supports, as we can then computationally test how changes in parameters at different levels would affect various performance indicators.

Besides covering theoretic foundations of ABMS, we focus heavily on hands-on learning as well. In particular, we will expose students to NetLogo, an intuitive yet powerful modeling language for building ABMS systems. We will be learning NetLogo by building several classical ABMS examples incrementally in class.

Objectives
Upon successful completion of this course, a student will be able to:
• Understand what is an ABMS.
• Evaluate the pros and cons of using an ABMS system in describing selected real-world phenomena.
• Utilize ABMS systems in policy/strategy evaluations.
• Appreciate the importance of considering uncertainty and opponent modeling when designing strategic, tactic, and operational policies.
• Complete the full cycle of building an ABMS system using the NetLogo programming language:
o Design an ABMS system with a proper level of granularity and fidelity (defining agents and means of communications).
o Validate and calibrate the built ABMS.
o Interpret the outcome of the ABMS system.

Course Learning Outcomes

1) Analyze real-world computational competition models.

2) Know tools availability for opponent modeling and analysis.

3) Explain the importance of opponent modeling in constructing ABMS.

4) Identify real-world scenarios that can benefit from using ABMS.

5) Identify the benefits and cost associated with the construction of ABMS.

6) Rigorously construct the framework for ABMS.

7) Understand feasible and reasonable mechanisms for individual actors to communicate with each other.

8) Decompose a complex business scenario into many actors representing physical entities in the scenario.

9) Implement ABMS for a number of real-world domains (e.g., finance, marketing, and operations management) using selected frameworks and tools (mostly NetLogo).

10) In a well-defined environment, design and construct major components that are associated to real-world roles (for example, various types of traders in financial markets and participants in a supply chain management scenario).

11) Know how to systematically explore the parameter space of an ABMS and conduct post-simulation analysis.

12) Effectively communicate and resolve conflicts while working in a self-chosen team with peers of various abilities and attitudes.

Discipline-Specific Competencies

Data Analytics, Applications Development, Computational Modelling, Data Visualisation, Intelligent Reasoning

SMU Graduate Learning Outcomes

Disciplinary Knowledge, Multidisciplinary Knowledge, Critical thinking & problem solving, Innovation and enterprising skills, Collaboration and leadership, Communication, Self-directed learning

Grading Basis

GRD - Graded