Workshop on Agentization (15-17 September 2021)

While many mathematical and statistical models in the social sciences consist of interacting agents, it is often the case that strong assumptions have to be made for reasons of analytical tractability (e.g., representative agents, rational agents, equal probability of interaction between agents, attainment of agent-level equilibrium).

Agent-based models (ABMs) are an emerging computational approach for studying social and natural phenomena in terms of interacting agents, and which facilitate the relaxation of unrealistic assumptions. Often ABMs address social phenomena about which other more conventional models exist, but direct comparisons of the output of the distinct models are not made directly or else are attempted only informally.

This workshop will focus on ABMs that reproduce the results of conventional models, and then generalize standard results by relaxing model specifications, usually in the direction of more realism. Such models agentize mathematical or econometric models and may demonstrate that conventional results are robust to certain parametric variations or are special cases of more general results.

The workshop on agentization will be held online through George Mason University, from 15-17 September, organized by the Computational Public Policy Lab and the Center on Social Complexity at Mason and sponsored by the Proteus Foundation. ABM pioneer W. Brian Arthur (Santa Fe Institute) will deliver a keynote address. We seek submissions of ABMs that closely reproduce conventional model results and then generalize them. Perhaps you have created an ABM that is similar to some standard model but which produces different results. If your ABM can be directly related to the standard model then it is of interest to this workshop. ABMs that have only notional relation to extant models are not suitable for this event.

People interested in presenting their research at this workshop should submit online (http://proteusfoundationseries.org/event/first-international-workshop-on-agentization-rendering-conventional-models-with-agent-based-computing/) by 16 August, either as a paper comparing results from an ABM to standard results, or as an abstract along with a working ABM or model output demonstrating the salience of an ABM to some existing model. It is our hope to combine a variety of papers describing agentized models from the workshop into an edited book or journal special issue. ABMs are welcome from any domain in which social processes are important. People interested in sitting in on or otherwise participating in this workshop are welcome to indicate their interest through the website. This is a free event.