Date of Award

Spring 5-2022

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Computational Modeling & Simulation Engineering

Program/Concentration

Modeling & Simulation Engineering

Committee Director

Andrew Collins

Committee Member

John Sokolowski

Committee Member

Hong Yang

Committee Member

Christopher Lynch

Abstract

Agent-based modeling (ABM) is a frequently used paradigm for social simulation; however, there is little evidence of its use in strategic coalition formations. There are few models that explore coalition formation and even fewer that validate their results against an expected outcome. Cooperative game theory is often used to study strategic coalition formation but solving games involving a significant number of agents is computationally intractable. However, there is a natural linkage between ABM and the study of strategic coalition formation. A foundational feature of ABM is the interaction of agents and their environment. Coalition formation is primarily the result of interactions between agents to form collective groups. The ABM paradigm provides a platform in which simple rules and interactions between agents can produce a macro level effect without large computational requirements.

This research proposes a hybrid model combining Agent-based modeling and cooperative game theory to find members of a cooperative game’s solution. The algorithm will be applied to the core solution of hedonic games. The core solution is the most common solution set. Hedonic games are a subset of cooperative games whereby agents’ utilities are defined solely by a preference relation over the coalitions of which they are members. The utility of an agent is non-transferrable; there can be no transfer, wholly or in part, of the utility of one agent to another. Determining the core of a hedonic game is NP-complete.

The heuristic algorithm utilizes the stochastic nature of ABM interactions to minimize computational complexity. The algorithm has seven coalition formation functions. Each function randomly selects agents to create new coalitions; if the new coalition improves the utility of the agents, it is incorporated into the coalition structure otherwise it is discarded. This approach reduces the computational requirements.

This work contributes to the modeling and simulation body of knowledge by providing researchers with a generalized ABM algorithm for forming strategic coalition structures. It provides an empirically validated model based on existing theory that utilizes sound mathematics to reduce the computational complexity and demonstrates the advantages of combining strategic, analytical models with Agent-based models for the study of coalition formation.

DOI

10.25777/g7vj-c853

ISBN

9798819394595

Share

COinS