EAGER: Demonstration of Scaling Impact on Coalition Formation in Agent-based Simulation

这个早期概念的探索性研究（EAGER）赠款奖将通过一系列专注于社会模拟的计算实验来推进建模和模拟领域。社会模拟被用来更好地理解国家面临的大规模，复杂的问题，如肥胖流行病和住房止赎危机。这项研究的重要性在于它有可能帮助人们更深入地了解这些复杂的问题，使政府官员在制定应对这些社会挑战的新政策时能够做出更明智的决定。具体来说，该奖项将研究个人行为如何随着模拟中活跃的决策代理数量而变化。通过与当地一所医学院的合作，该赠款将有助于深入了解与医院医生越来越多的安置有关的问题。该基金还将在弗吉尼亚州东南部的少数民族服务机构Old自治领大学（Old Dominion University）开设一门关于多个决策者的情景建模的跨学科课程。该基金将通过设计和分析一系列基于计算主体的仿真建模实验，挑战人们普遍认为的行为模型与所研究的社会群体的规模无关的假设。具体来说，研究的目标是证明规模的联盟形成行为在基于代理的模拟的影响。一个博弈理论框架将用于开发模拟的概念模型、建模场景和测试数据集。实验涉及使用不同规模的机器学习开发行为子模型。计算实验将针对各种场景、游戏类型和适应度函数进行重复。本研究假设，衍生的行为子模型将以可解释的方式在不同规模之间存在显着差异。该奖项反映了NSF的法定使命，并且通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

This EArly-concept Grants for Exploratory Research (EAGER) award will advance the field of Modeling and Simulation through a series of computational experiments focused on social simulation. Social simulations are used to understand better the large-scale, complex problems that the nation faces such as the obesity epidemic and the housing foreclosure crisis. The significance of this research lies in its potential to help provide a deeper understanding of these complex issues, enabling government officials to make more informed decisions when formulating new policies to tackle these societal challenges. Specifically, this award will examine how individual behavior changes with the number of decision-making agents active in the simulation. Through a partnership with a local medical school, the grant will help provide insight into the issues relating to the increasing placement of doctors in hospitals. This grant is accompanied by an educational plan to develop a cross-disciplinary course on modeling situations with multiple decision-makers at Old Dominion University, a minority-serving institution in southeastern Virginia.This grant will challenge the widespread assumption that behavioral models remain invariant regardless of the size of the social community under study by designing and analyzing a series of computational agent-based simulation modeling experiments. Specifically, the research goal is to demonstrate the impact of scale on coalition formation behavior within an agent-based simulation. A game-theoretical framework will be used for the development of the simulation’s conceptual model, the modeling scenario, and the test dataset. The experiments involve developing behavioral sub-models using machine learning at different scales. The computational experiments will be repeated for various scenarios, game types, and fitness functions. This research hypothesizes that the derived behavioral sub-models will differ substantially between scales in an explainable manner.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.