Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

TD38

agents there. Each agent is free to move between locations, and at each time decides whether to stay at the same location or switch to another one. We study the equilibrium behavior of the agents as a function of dynamics of the stochastic resource process and the nature of resource sharing in the limit where the number of agents and locations increase proportionally. 2 - Mean Field Equilibrium: Uniqueness, Existence, Comparative Statics, and Applications Bar Light, Stanford GSB, Stanford, CA, United States, Gabriel Weintraub The standard solution concept for stochastic games is Markov perfect equilibria (MPE), but the computation of MPE becomes intractable as the number of players increases. We instead consider mean field equilibria (MFE) that have been popularized in the recent literature. We make three main contributions. First, our main result in the paper provides conditions that ensure the uniqueness of an MFE. Second, we generalize previous existence results of an MFE. Third, we provide general comparative statics results. We apply our results to dynamic oligopoly models commonly used in previous work. We believe our uniqueness result is the first of its nature in the class of models we study. 3 - The Cutoff Structure of Top Trading Cycles in School Choice Irene Yuan Lo, Columbia University, New York, NY, United States The prominent Top Trading Cycles (TTC) mechanism has attractive properties for school choice, as it is strategy-proof and Pareto efficient. However, these properties are obfuscated by the common combinatorial description of TTC. We show that the TTC assignment can be described by admission cutoffs for each pair of schools. In a large-scale continuum model these cutoffs can be computed directly from the distribution of preferences and priorities, providing a framework for evaluating policy choices. As an application of the model we solve for optimal investment in school quality under choice and find that an egalitarian distribution can be more efficient as it promotes better sorting by students. 4 - The Ubiquity of Aggregate Uncertainty in Mean Field Equilibria Daniel Lacker, Columbia University, 500 W. 120th St, Mudd 306, NY, United States A game with a continuum of agents is typically justified by showing that, if the mean field (continuum) game has a unique equilibrium, then any sequence of equilibria in the corresponding N-player games converges as N goes to infinity to this mean field equilibrium (MFE). For static games, even when the MFE is not unique, limit points of N-player games can still be characterized as mixtures (randomizations) over the set of MFE. However, in dynamic games with nonunique MFE, there are other randomized limit points which cannot be described as mixtures of MFE; the relevant aggregate quantities evolve stochastically. The goal of this talk is to explain this phenomenon and provide examples. n TD38 North Bldg 225B Design Innovations in Matching Markets and Systems Sponsored: Applied Probability Sponsored Session Chair: Vahideh Manshadi, Yale University, New Haven, CT, 06511, United States 1 - Carta: A Data-driven Course Planning Platform Ramesh Johari, Stanford University, Huang Engineering Center Rm 311, 475 Via Ortega, Stanford, CA, 94305-4121, United States, Sorathan Chaturapruek, Thomas Dee, Rene Kizilcec, Mitchell Stevens College students rely on increasingly data-rich environments when making learning-relevant decisions about the courses they take and their expected time commitments. In this talk we discuss recent research involving Carta, a course exploration platform for undergraduates. In particular, we discuss two field experiments that investigate the consequences of such a platform on college students’ GPA. 2 - On the Optimal Design of a Bipartite Matching System Rene A. Caldentey, The University of Chicago, Booth School of Business, 5807 S. Woodlawn Ave, Chicago, IL, 60637, United States, Philipp Afeche, Varun Gupta In this talk, we explore the optimal design of matching topologies for a multi-class multi-server queueing system in which each customer class has a specific preference over server types. We investigate the performance of the system from the perspective of a central planner who must decide the set of feasible customer- server pairs that can be matched together under fairness constraints for both customers and servers.

n TD36 North Bldg 224B Robotics, Drones, and AVs in Logistics Emerging Topic: Robotics, Drones and Autonomous Vehicles in Logistics Emerging Topic Session Chair: Michael Levin 1 - Unmanned Aerial Vehicle Path Planning for Network State Estimation Cesar N. Yahia, The University of Texas at Austin, Austin, TX, 78705, United States, Stephen D. Boyles, Christian G. Claudel We investigate the problem of planning a trajectory for an unmanned aerial vehicle with the objective of estimating network state parameters. The optimal trajectory would maximize the information on the network state over a finite time horizon. In this case, planning and estimation are coupled since different paths chosen by the mobile sensor generate distinct observations, which in turn directly influences the quality of the estimates. To address this problem, we propose an online control algorithm based on forward propagation of traffic flow. 2 - Traffic Optimization for a Mixture of Self-interested and Compliant Agents Michael Albert, University of Virginia, Charlottesville, VA, United States, Guni Sharon, Peter Stone, Stephen D. Boyles, Tarun Rambha This paper focuses on two commonly used path assignment policies for agents traversing a congested network: selfinterested routing, and system-optimum routing. A computationally tractable method is presented that computes the minimal amount of agents that the system manager needs to influence (compliant agents) in order to achieve system optimal performance. Moreover, this methodology can also determine whether a given set of compliant agents is sufficient to achieve system optimum and compute the optimal route assignment for the compliant agents to do so. Experimental results are presented also presented. 3 - Who Should Drive? Optimal Switching Policy Between Driving Entities in Semi-Autonomous Vehicles Franco van Wyk, University of Tennessee, Knoxville, TN, United States, Anahita Khojandi, Neda Masoud Autonomous vehicles are expected to increase safety on the road. However, currently automated entities do not necessarily outperform human drivers under all circumstances. Therefore, in certain conditions it is safer for the human driver to take over the control of the vehicle. However, switching control back and forth between the driver and the automated driving-entity may itself pose a short-term, elevated risk. We develop a Markov decision process model to determine the optimal driving-entity switching policy that minimizes the expected safety risk of a trip, considering the dynamic changes of the road/environment during the trip. We conduct numerical analyses and provide insights. 4 - Leveraging Shared Autonomous Electric Vehicles for First/last Mile Mobility T. Donna Chen, Assistant Professor, University of Virginia, P.O. Box 400742, Charlottesville, VA, 22904, United States, Farhan Javed A simulation framework is proposed to evaluate shared autonomous electric vehicle (SAEV) operations that center around Tukwila light rail station to provide first/last-mile mobility in Seattle. Results show great potential for leveraging SAEVs to increasing the station’s catchment area, resulting in reduced parking demand. The proposed SAEV fleet reduce system-wide VMT by 36.65% through ridesharing. Fast charging technology decreases the fleet size and wait time by 56% and 72.97%, respectively. Decision makers can apply this framework to evaluate the service by comparing scenarios with varying vehicle configuration, capacity, and charging infrastructure. n TD37 North Bldg 225A Large-Scale Games Sponsored: Applied Probability Sponsored Session Chair: Gabriel Weintraub, Stanford Graduate School of Business, Stanford, CA, 94304, United States 1 - Mean Field Equilibria for Resource Competition in Spatial Settings Krishnamurthy Iyer, Cornell University, Ithaca, NY, 14850, United States, Pu Yang, Peter Frazier Inspired by crowd-sourced transportation services and other location-based activities, we consider a model of competition among nomadic agents for time- varying and location-specific resources. Each agent derives a periodic reward based on the overall resource level at her location, and the number of other

359