Algorithmic model A model specified as a sequence of procedures that follow one another, where the exact sequence depends on the outcome of each procedure. Algorithmic models are often described using flowcharts or language constructs such as ‘if X then do Y else do Z’, where X is some condition and Y and Z are procedures.
Agent-based model A computer simulation model which implements a set of autonomous agents situated in an artificial environment. Each agent pursues its own goal(s), typically adjusting its behavior according to its current state and the environment in which it finds itself. In archaeological applications agents usually represent individuals or households, but they can theoretically represent any relevant unit ranging from individual atoms or molecules to entire nation states.
Artificial intelligence (AI) The field concerned with the construction of machines (i. e., artificial devices) that can reason and solve problems. Proponents of ‘strong’ AI seek to reproduce human intelligence, while those operating under the banner of ‘weak’ AI do not believe that this is possible and/or are content to focus on more narrowly defined problem domains. artificial life (AL) The field that studies the systems responsible for the maintenance and evolution of life by attempting to replicate them, usually, but not exclusively, in a computer simulation. Proponents of ‘strong’ AL hold that life is substrate neutral, from which it follows that the life in an AL simulation could potentially be as real as biological life, while proponents of the ‘weak’ position hold that true life is confined to ‘wet’ chemistry. cellular automata In practice, a computer simulation which iteratively (and synchronously) updates the state of each of a set of identical cells arranged on a regular grid. Each cell may be in one of a finite number of states and has an update-rule which specifies how it should change its state depending on the states of other cells in its neighborhood. cognitive architecture In the context of agent-based modeling, a formal representation of the process by which an agent reasons. In practice this often amounts to a representation of how the agent decides what to do in order to achieve its goal(s). The best-known example is based on Michael Bratman’s ‘belief-desire-intention’ theory of human practical reasoning.
Dynamical systems model A model specified in terms of one or more mathematical equations which deterministically predict how one or more quantities (state variables) change through time, given their initial values. Many such models are capable of producing very different behaviors, ranging from monotonic progression to an equilibrium value (e. g., smooth growth to a stable population size) to chaotic oscillations (e. g., a population size that undergoes a never exactly repeated sequence of growth and decline).
Expert system A reasoning system in which a computer program (the inference engine) attempts to solve a problem by applying rules stored in a domain specific knowledge base to data also stored in that knowledge base. object-oriented programming language A computer programming language, such as C++ or Java, that allows data and operations to be kept together in objects. This approach has many advantages from a software engineering perspective, but the principal interest here is that object-oriented programming lends itself particularly well to the implementation of agent-based models, since agents are readily created as objects.
Sensitivity analysis The process of re-running a computer simulation model with many different combinations of parameter values in order to investigate: (1) the range of results that are possible and (2) which parameters have a significant effect on the results.
Stochastic A stochastic model incorporates one or more random processes. A stochastic computer simulation model should be run many times, each with a different seed for the random number generator, in order to determine the extent to which the results vary by chance alone. This procedure should be followed for each combination of parameter values studied. toolkit A collection of computer program codes, typically organized as libraries, that provides much of the functionality required to implement a particular type of computer program. For example, the Repast simulation toolkit provides most of the generic features necessary for an agent-based model, thus
Allowing the programmer to concentrate on implementing features specific to the problem at hand. validation The process of ensuring that a computer simulation model reflects the salient aspects of the real-world phenomenon that is being studied. Ideally this is achieved by demonstrating that the model can reproduce observed outcomes from known initial conditions, but this is often difficult or impossible in archaeological applications of simulation. Consequently, archaeological models are often validated using sensitivity analysis to demonstrate that model outcomes are theoretically reasonable for a wide range of parameter values. It is vital that any archaeological data that one wishes to explain using the model is not also used for validation. verification The process of ensuring that the computer simulation model correctly implements the conceptual model, in other words, that there are no errors in the program code.
World History
