|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
TERMS: MULTI-AGENT COMPUTER MODELS With the increasing power of computers over the past decade, it is now possible to create elaborate computer models of agents and their interactions. Various terms have been used to describe this approach, such as multi-agent models, bottom-up modeling, artificial life or artificial social systems, and so on. The basic purpose of these models is to help people understand the properties of complex biological or social systems through the analysis of the behavior of the computer simulations. Multi-agent models are typically used in different ways than traditional computer models. Robert Axelrod suggests that these models provide a way of doing thought experiments. By very carefully and rigorously specifying a set of rules about a system one wishes to better understand, one can develop intuitions about how the system might behave under different conditions. Due to the multiplicity of ways agents might interact, nonlinearities and differing initial conditions, evolution and adaptive processes, and random factors, executing a multi-agent model multiple times will almost never run the same way twice. However, one generally finds very interesting patterns (i.e., self-organization) that emerge under certain rule conditions one might specify for how the agents in the model interact. It has been said the use of multi-agent computer modeling is a third way of doing science, complementing the use of direct observations and the use of mathematics. We are just now starting to see the use of multi-agent models in the social sciences and now in the business world, with initial applications probably in areas such as planning and decision-making. |
|
|
|||||||
home | about us | careers | client login | business needs | real resources | real world | enterprise evolution ©2002 LoBue | site map | site feedback |
|||||||