Individual-Based Models

an annotated list of links
by Craig Reynolds



Individual-based models are simulations based on the global consequences of local interactions of members of a population. These individuals might represent plants and animals in ecosystems, vehicles in traffic, people in crowds, or autonomous characters in animation and games. These models typically consist of an environment or framework in which the interactions occur and some number of individuals defined in terms of their behaviors (procedural rules) and characteristic parameters. In an individual-based model, the characteristics of each individual are tracked through time. This stands in contrast to modeling techniques where the characteristics of the population are averaged together and the model attempts to simulate changes in these averaged characteristics for the whole population. Individual-based models are also known as entity or agent based models, and as individual/entity/agent-based simulations.

Some individual-based models are also spatially explicit meaning that the individuals are associated with a location in geometrical space. Some spatially explicit individual-based models also exhibit mobility, where the individuals can move around their environment. This would be a natural model, for example, of an animal in an ecological simulation. Whereas plants in the same simulation would not be mobile. Some individual-based models are not spatially explicit, for example a simulation of a computer network might be based on individual models of the networked computers, but their location would be irrelevant. Spatially explicit models may use either continuous (real valued) or discrete (integer valued, grid-like) space.

Individual-based models are a subset of multi-agent systems which includes any computational system whose design is fundamentally composed of a collection of interacting parts. For example an "expert system" might be composed of many distinct bits of advice which interact to produce a solution. Individual-based models are distinguished by the fact that each "agent" corresponds to autonomous individual in the simulated domain.

There is an overlap between individual-based models and cellular automata. Certainly cellular automata are similar to spatially-explicit, grid-based, immobile individual-based models. However CAs are always homogeneous and dense (all cells are identical), whereas a grid-based individual-based model might occupy only a few grid cells, and more than one distinct type of individual might live on the same grid. (Of course a CA can have cells in various states, and so represent concepts like empty or occupied by type 3. Perhaps the significant difference is whether the simulation's inner loop proceeds cell by cell, or individual by individual. (Although that distinction is muddied by parallel-processing hardware.)) The philosophical issue is whether the simulation is based on a dense and uniform dissection of the space (as in a CA), or based on specific individuals distributed within the space.

Of course, note that everyone uses terminology differently, so take the definitions above with a grain of salt. ("Your mileage may differ.")

My interest in this area began when I made a model of bird flocks and related group motion. As a result I am particularly interested in individual-based models using spatially explicit mobile agents in continuous space. This bias may be reflected in the selection of resources listed below.


Online resources

These are general purpose software toolkits useful for implementing individual-based models. Listed below are applications of individual-based models, arranged by general topic area.

Offline resources


Laboratories and Groups


Journals


Conferences


Send comments to Craig Reynolds <cwr@red.com>
5645 visitors since September 28, 1997
Last update: August 2, 1999