2005 World Conference on Natural Resource Modeling

Arcata, California, June 14-17 2005

Special workshop

The conference will be followed by a one-day (June 18) workshop on individual-based modeling. Individual-based models (IBMs) are simulation models that represent a system’s dynamics as emerging from the behavior of the system’s individuals as they interact with each other and their environment. IBMs have many potential advantages for modeling populations and communities: environmental effects can be represented mechanistically and explicitly, adaptive behaviors (e.g., "trait-mediated interactions") and their consequences can be modeled, and processes acting at different (or across) spatial and temporal scales and ecological levels (e.g., individual- vs. population-level) are easily accommodated. IBMs make a wide range of new ecological and resource management problems amenable to modeling analysis.

IBMs are now widely used, but they pose unique challenges: IBMs can be very complex and difficult to understand, they often appear highly ad-hoc with little relation to theory, and computer implementation approaches that work for simpler models often doom an IBM to failure.

The workshop will be organized around three major challenges in building and using IBMs.

  • Complexity: How can a model be designed so it is complex and realistic enough to solve important, real problems, while otherwise being as simple as possible? What techniques can determine which variables and processes should be included vs. excluded?
  • Theoretical and conceptual bases: The essential characteristics of most IBMs cannot be described well using either traditional mathematics or conventional ecological theory. What common conceptual basis can we use instead to design and describe IBMs? What is "theory" in models that link individual and system levels? How can this theory be developed and applied?
  • Software: For IBMs, software must do more than implement the model; it must also provide a laboratory in which we can observe and conduct experiments on the virtual population in the model. Software development can be a much greater effort for IBMs, but there are specialized platforms that are very helpful. What is the appropriate platform for a particular IBM, and how can modelers get started using it? The workshop will introduce NetLogo, a widely used, high-level platform appropriate for IBMs of moderate complexity and prototypes. Also introduced will be Swarm and RePast, which are essentially programming languages specifically for IBMs. Recommendations for selecting and learning to use these platforms will be provided.

The workshop will include exercises implementing models in the NetLogo and Repast software platforms. Computers with these packages installed will be available for participants, but people who wish to use their own computers are encouraged to install the software in advance.

NetLogo 2.1 is available at:
http://ccl.northwestern.edu/netlogo/download.shtml

The Repast Suite is available at:
http://repast.sourceforge.net/download.html

Preliminary Agenda

9:00 - Introductions

9:15 - What are IBMs? Why use IBMs? What are the benefits and problems of using IBMs?

9:30 - A conceptual framework for IBMs: How do we describe and design models that are not described well by traditional mathematics?

10:30 - Break

10:45 - Getting started: How to design an IBM so it is "as simple as possible, but not simpler".

11:00 - Introduction to the NetLogo modeling platform. NetLogo provides a simple programming language and graphical interface for simple IBMs and prototypes.

12:00 - Lunch

13:00 - Analyzing IBMs: What to do once your model runs. How do you do theoretical or management science with an IBM?

13:30 - Software for IBMs: Why is software a big deal? What tools are available? How do you adequately test an IBM's software?

14:00 - Introduction to Repast, a widely used programming environment for
complex IBMs. See the tutorial summary below.

16:00 - Wrapup

16:30 - Change of habitat and biochemically facilitated discussion.

The workshop is organized by Steve Railsback (Lang, Railsback & Assoc. and Humboldt State University). Dr. Railsback has over 10 years experience managing and conducting development of IBMs and their software, and applying IBMs to natural resource management issues. He co-authored the new book Individual-based Modeling and Ecology, which provides extensive treatment of the topics covered in the workshop.

Other participants will include a representative of the Repast development team at Argonne National Laboratory; and Andrew Kanarek and Steve Jackson (current and former students in the Humboldt State environmental modeling program who are experienced users of the NetLogo and Swarm software platforms).

Agent-Based Modeling and Simulation with Repast

The REcursive Porous Agent Simulation Toolkit (Repast) is a free, open source agent-based modeling and simulation library with pure Java and pure C# implementations. Repast seeks to support the development of extremely flexible models of agents with special support for representing social interactions. Users build simulations by incorporating Repast library components into their own programs. Repast is being developed and maintained by the Repast Organization for Architecture and Design (ROAD). More information on Repast, as well as free downloads, can be found at the Repast home page, repast.sourceforge.net . The tutorial will introduce the use of Repast for agent modeling. The outline of the tutorial is as follows:

  • The background of Repast will be discussed in the context of agent-based modeling and simulation.
  • The core Repast components will be discussed:
    • The Repast simulation control system will be detailed.
    • The Repast scheduler will be described.
    • The Repast support tools will be introduced with an emphasis on behavioral components.
  • An example Repast simulation will be reviewed.