PhD Thesis Proposal: Reed Harder

Monday, July 22, 2019, 4:00–6:00pm

Jackson Conference Room, Cummings Hall

“Competitive Resource Allocation over a Network: Optimization and Inference in Air Transportation Systems and Beyond”


Engineering systems characterized by multiple resource-allocating agents are major components of critical industrial domains, including transportation, energy distribution, and facility location. This dissertation aims to develop tools for the analysis of such systems, where multiple competing agents aim to optimally allocate resources over a network. The analysis of these systems, both predictive and prescriptive, presents multiple challenges that are both analytical and computational in nature. Optimization problems for the optimal allocation of resources across networks for single agents can be challenging enough on their own. However, in many systems of interest, inputs into these optimization problems are often related to the behaviors of other optimizing agents. This makes prediction and decision-making particularly challenging, both for individual agents and system regulators.

We analyze competition in the air transportation system as an exemplar of such a system, and draw connections to related problems in related domains. We begin with an analysis of individual agent service location decisions. These decisions can be temporal, as in the scheduling of a flight, or spatial, as in the location of a retail store. We develop optimization and game theoretic models of such decisions in competitive scenarios to analyze the impact of different pricing strategies on location incentives, and quantify the relationship between location decisions and higher-level aggregate service frequencies (e.g., the number of flights per day between various airports) across a network. Forecasting models integrating the structure of these networks and the higher-level service frequencies deployed on them are developed for the predictive modeling of airline network evolution. We then analyze and calibrate a game theoretic model of airline frequency and pricing decisions, and evaluate this model’s use in prediction and scenario analysis in an airline network. Finally, we will extend this game theoretic model to more complex networks, where high levels of passenger connectivity impose strong couplings between frequency allocation decisions.

Thesis Committee

For more information, contact Daryl Laware at