2016 INFORMS Annual Meeting Program

TD26

INFORMS Nashville – 2016

3 - V-shaped Sampling Based on Kendall-distance To Enhance Optimization With Ranks Haobin Li, Scientist, A*STAR, Institute of High Performance Computing, 1 Fusionopolis Way, #16-16 Connexis North, Singapore, 138632, Singapore, lihb@ihpc.a-star.edu.sg, Giulia Pedrielli, Loo Hay Lee, Ek Peng Chew, Chun-Hung Chen Optimization over rank values has been of concern in multi-fidelity simulation optimization. Specifically, Chen et al. (2015) proposes the concept of Ordinal Transformation (OT) to translate multi-dimensional discrete optimization problems into simpler single-dimension problem, where the dimension being used is the rank in the ordinal space. In this paper we build on the idea of OT in order to derive an efficient sampling algorithm to identify the solution with the best rank in the settings of multi-fidelity optimization. We refer to this algorithm as V-shaped, in which the concept of Kendall distance adopted in the machine learning theory, is used to characterize solutions in the OT space. 4 - Bicycle-sharing With Reallocation Trucks And Private Exchange Taipei YouBike System as Example Hui-Chih Hung, National Chiao Tung University, Hsin-Chu, 300, Taiwan, hhc@nctu.edu.tw, Jun-Min Wei, Ming-Te Chen Subject to limited numbers of bicycles and docks, we consider trucks for bicycle reallocation and mobile apps for private exchange in bicycle-sharing systems. Trucks are hired to dynamically redistribute bicycles among unbalanced stations and mobile apps are used to transfer bicycles among users without docks. This allows bicycle exchange even when all docks are full. Three objectives are studied: (1) maximizing the utilization of bicycles, (2) maximizing the net profit of system, and (3) optimizing the fleet sizes of bicycles and trucks. Finally, mathematical programming models are built and real data of Taipei YouBike system from 2013 to 2015 are adopted for numerical study. Invited: Auctions Invited Session Chair: Oleg Baranov, University of Colorado at Boulder, 256 UBC, University of Colorado, Boulder, CO, 80309, United States, oleg.baranov@colorado.edu 1 - Efficient Dynamic Auction For U-shaped Returns Oleg Baranov, Colorado, Oleg.Baranov@Colorado.edu When bidders have decreasing returns, the efficient dynamic auction is well- known. Recently, Baranov et al. (2016) described an efficient dynamic auction for bidders with increasing returns. In this paper, we design an efficient auction for bidders with single-peaked returns. For auctions to buy, our setting includes one of the most typical cost structures in economics. For auctions to sell, our setting is a good approximation for single-band spectrum auctions. 2 - Obtaining The Final Channel Assignment In The Federal Communications Commission’s First-ever Incentive Auction Karla L Hoffman, Professor, George Mason University, Mail Stop 4A6, 4400 University Drive, Fairfax, VA, 20124, United States, khoffman@gmu.edu, Brian Smith, Steven Charbonneau, James Costa, Tony Coudert, Rudy Sultana In this talk, we will discuss the procedure for determining the Final Channel Assignment for U.S. and Canadian broadcasters at the conclusion of the “Incentive Auction.” The FCC is utilizing a sequence of optimizations to create a channel assignment that will be the least disruptive to both broadcasters and the over the air television viewers. We will outline this sequence and explain how this sequence satisfies the objectives of the FCC, Industry Canada and broadcasters. 3 - Determining The Stations Not Needed In The Federal Communication Commission’S First-ever Incentive Auction Karla L Hoffman, George Mason University, System Eng and Operations Research Dept, 4400 University Drive Mailstop 4a6, Fairfax, VA, 22030, United States, khoffman@gmu.edu, James Andrew Costa, Steven Charbonneau, Anthony Coudert, Brian Smith, Rudy K Sultana The FCC uses a novel descending-price auction to determine the spectrum to be purchased from broadcasters. The auction is designed for stations to compete until there are no channels available in the market. If a station always has a channel available, they are not needed in the auction. An optimization procedure was used to determine whether a station always had a channel available, and therefore had no chance of winning in the auction. TD26 110B-MCC Spectrum Auction

4 - Combinatorial Land Assembly Tzu-Yao Lin, University of Maryland, LinT@econ.umd.edu We propose a reverse auction for real estate developers to acquire complementary urban lands from multiple owners. Apart from the all-or-nothing mechanisms in the previous literature, we determine the set of land parcels to be assembled in a descending clock auction, which gradually lowers the offer to each remaining owner until the trading condition is met. This mechanism is obviously strategyproof for sellers. The optimal price adjustment trajectory is a solution for the corresponding stochastic optimal control problem, which minimizes the expected welfare loss from inefficient rejection.” TD27 201A-MCC Stochastic Modeling In Healthcare Operations Sponsored: Manufacturing & Service Oper Mgmt Sponsored Session Chair: Carri Chan, Columbia Business School, New York, NY, United States, cwchan@columbia.edu Co-Chair: Vahid Sarhangian, Columbia Business School, New York, NY, United States, vs2573@columbia.edu 1 - Identify Optimal Overflow Policies Using Approximate Dynamic Programming Pengyi Shi, Purdue University, shi178@purdue.edu, Jim Dai To alleviate Emergency Department congestion, boarding patients who wait to be admitted to inpatient wards may have to be overflowed to a non-primary ward when they wait too long. We develop approximate dynamic programming tools to identify the optimal overflow policies under different system states. 2 - Yardstick Competition For Emergency Department Queues Ozlem Yildiz, University of Rochester, Rochester, NY, United States, ozlem.yildiz@simon.rochester.edu, Nicos Savva, Tolga Tezcan We study whether an alternate pay-for-performance method can alleviate ED overcrowding through incentivizing socially-desired ED capacity levels, although the healthcare regulator does not know the capacity cost structure. Using yardstick competition, we propose a regulatory scheme that achieves this using the wait time and arrival rate information of each ED. 3 - Timing Of Hospital Discharges Matters Jonathan Helm, Indiana University, helmj@indiana.edu, Rene Bekker The mismatch in timing of arrivals and discharge processing in hospitals leads to a census process that causes the hospital to experience significant congestion in the middle of the day. This leads to a chaotic environment and major operational efficiencies. In this research we formulate and analyze a stochastic census process to investigate the effect of the timing of doctor’s discharge processing on inpatient census levels and identify new approaches to discharge processing that can alleviate congestion and also provide benefits to the patients being discharged themselves. 4 - Dynamic Server Allocation In A Multiclass Queueing System With Shifts: Nurse Staffing In Emergency Departments Vahid Sarhangian, Columbia Business School, New York, NY, United States, vs2573@columbia.edu, Carri Chan Nurse staffing decisions in emergency departments (EDs) are typically assigned weeks in advanced, which can create staffing imbalances as patient demand fluctuates. In this work, we consider the potential benefits of assigning nurses to different areas within an ED at the beginning of each shift. We study the problem of optimal reassignment of nurses to areas by considering a multiclass queueing model of the system. We analyze an associated fluid control problem and use the solution to develop policies that achieve asymptotically optimal performance under fluid-scaling for the original stochastic system. We find this additional flexibility can substantially reduce waiting times for patients.

342