An Environmental Modeling Language for Agents and Fields

Environmental modeling involves manipulating environmental attributes represented in software by agents, fields or both, but most modeling environments are designed to be especially useful for either agent-based or field-based modeling. Agent-based and field-based modeling environments have different properties with respect to their ease of use and how well both agents and fields can be represented and manipulated. Most agent-based modeling environments require the modeler to use a general purpose object oriented programming language like Java to express models, while field-based modeling environments often implement a high level, domain specific imperative language based on map algebra, or extent a general purpose scripting language. Because of this, field-based models are more easily defined by domain experts than agent-based models. On the other hand, because a lower level language is used, agent-based modeling environments are more easily extended by missing functionality, like support for fields, while in general, field-based modeling environments lack support for defining agents. We are working on a new environmental modeling environment which is designed from the ground up for manipulating both agents and fields. Our goal is to combine the advantages of current agent-based and field-based modeling environments. For this, a conceptual data model is developed which is capable of storing both agent and field data, and which is a superset of the traditionally separate data models for agents and fields. Given this data model a high level domain specific imperative language is designed which enables domain experts to define combined agentand field-based models.