Growing artificial societies: social science from the bottom up
Growing artificial societies: social science from the bottom up
Cellular-automata models applied to natural hazards
Computing in Science and Engineering
| Hi-index | 0.00 |
This paper describes a technique for determining which systems should be modeled in higher detail when modeling Very Large and Complex Systems (VLCS) using Agent-based Modeling techniques. The goal of this research was to develop a method that (1) could be applied in the conceptual modeling phases (i.e., before the bulk of the modeling effort began), and that (2) would produce quantitative recommendations as to which agents composing the VLCS should be modeled in more detail. The technique described relies on studying the functional interactions between the systems composing the VLCS, i.e., how the entities in the VLCS interact with one another. Depending on the properties of the networks formed by these interactions, different recommendations arise as to how the modeling effort should be distributed amongst the entities. Fourteen different network node metrics were studied and tested. The results indicate that when the behavior of the network is not chaotic, the Perron-Frobenious Eigenvector and the Fiedler Eigenvector provide the best measures for identifying the criticality of each node in the network. When the network topology displays certain characteristics which can be associated with it producing chaotic behaviors, it is best to equally distribute the modeling effort amongst all the nodes. This result seems intuitive since in chaotic systems small variations can produce radically different behavior trajectories. The technique was tested using Random Boolean Networks and Agent-based Simulations.