2016 INFORMS Annual Meeting Program

POSTER COMPETITION

INFORMS Nashville – 2016

Two-stage Game Theoretic Modeling Of Airline Frequency And Fare Competition Reed Harder, Dartmouth College, 14 Engineering Drive, Hanover, NH, 03755, United States, reed.haseltine.harder.TH@dartmouth.edu, Vikrant Vaze We develop a 2-stage game-theoretic model of airline competition, with flight frequency decisions followed by fare decisions. For a simple 2-player model, we show that this game has properties that ensure a tractable and credible equilibrium solution. We then use quadratic approximations of payoffs to extend these results to more realistic scenarios. Finally, we calibrate model parameters on a real-world network and validate out-of-sample predicted frequencies against observed airline behavior. Identification And Allocation Of Increased-risk Encephalitis Organs Pinar Keskinocak, Georgia Institute of Technology, School of Indust/System Eng, 765 Ferst Drive, Atlanta, GA, 30332, United States, pinar@isye.gatech.edu, Hannah Smalley, Nishi Anand, Dylan Buczek, Nicholas Buczek, Tim Lin, Tanay Rajore, Muriel Wacker, Brian Gurbaxani, Matthew Kuehnert, Sridhar Basavaraju, Teresa Hammett We developed decision-support tools to aid organ transplant physicians (and patients) in the identification and allocation of organs that carry the risk of infectious encephalitis. The Infectious Encephalitis Risk Calculator assesses whether a donor (and his/her organs) may have infectious versus non-infectious encephalitis. The Liver Transplant Decision Aid helps patients and physicians evaluate the trade-offs between accepting and rejecting an increased-risk encephalitis liver, thus potentially enabling a better allocation of high-risk organs and reducing deaths on the waitlist. The tool provides wait time estimates for liver transplants based on patient characteristics. Evaluation Of Google’s Voice Recognition And Sentence Classification For Health Care Applications M. Majbah Uddin, University of South Carolina, 300 Main Street, Civil and Environmental Engineering, Columbia, SC, 29208, United States, muddin@cec.sc.edu, Nathan Huynh, Jose M Vidal, Kevin M Taaffe, Lawrence Fredendall, Joel S Greenstein This study examined the use of voice recognition (VR) technology in perioperative services (Periop) to enable Periop staff to record workflow milestones using mobile technology. The use of mobile technology to improve patient flow and quality of care could be facilitated if such VR technology could be made robust. The goal of this experiment was to allow the Periop staff to provide care without being interrupted with data entry and querying tasks. This study enhanced Google’s VR capability by using post-processing classifiers (i.e., bag-of-sentences, support vector machine, and maximum entropy), which would facilitate its use in health care and other applications. A Heuristic Algorithm Assigning Optimal Tolls Vyacheslav V. Kalashnikov, Tecnologico de Monterrey, ITESM, Campus Monterrey, Ave. Eugenio Garza Sada 2501 Sur, The problem of assigning optimal tolls to the arcs of a multi-commodity transportation network is formulated as a bilevel mathematical program. We describe an algorithm based on the allowable ranges to stay optimal (ARSO) resulting from sensitivity analysis applied to the lower level problem. In this way, one can analyze possible changes in the coefficients of some variables in the objective function within these allowed ranges without affecting the optimal solution. In addition, when stuck to a local maximum solution, the “filled function” technique helps one “jump” to another local maximum (if such does exist) or stop the search. Numerical experiments confirm the robustness of our heuristics. Developing A Novel Inventory Classification Approach For Large Scale Multi Echelon Inventory Systems Monterrey, 64849, Mexico, kalash@itesm.mx, Nataliya I. Kalashnykova, Arturo García-Martínez The purpose of this research is to create a new inventory classification approach for large-scale multi-echelon repairable item provisioning systems. In this research, we develop a new concept for defining the classification criteria which is the artificial stocking policy (ASP). We also propose a new partitioning approach which takes into the account the characteristics of the (aggregated) pseudo-items. The proposed technique is evaluated and compared with complete enumeration and eight alternative clustering and classification methods via 36 different problem instances. The results indicate that the proposed methods significantly outperform the alternative techniques. Alireza Sheikhzadeh, University of Arkansas, 4207 Bell Engineering Center, Fayetteville, AR, 72701, United States, asheikhz@uark.edu, Manuel D Rossetti

The Value Of Aggregation Under Minimax Pricing Scheme In The Electricity Retail Market Alberto J Lamadrid, Associate Professor, Lehigh University, 621 Taylor Street, R451, Bethlehem, PA, 18015-3120, United States, ajl259@cornell.edu, Kwami Senam Sedzro, Mooi Choo Chuah We explore both economic and technical benefits of demand responsive load aggregation under a specific retail pricing scheme we call Minimax. It is a 3-rate scheme with a threshold level. Each region is applied a differentiated rate. To assess how adopting Minimax would impact both end-users and distribution system operators (DSO), we model consumers’ optimal response to Minimax as an MILP and evaluated the DSO’s operating conditions and costs. A case study with different aggregation scenarios implemented on the IEEE 33-bus system reveals that larger aggregate groups achieve lower energy bills and help the DSO lower generation cost and aggregate peak demand, and achieve better voltage profiles. Designing A Space-efficient Warehouse Layout In this research, we analyze the factors that impact the space utilization in a warehouse. We develop mathematical models to maximize space utilization in the warehouse and depict the trade-off exist between space utilization and transportation cost in the warehouse layout. Optimizing Decisions In Prenatal Integrated Screening For Down Syndrome Jia Yan, Georgia Institute of Technology, 710 Peachtree Street NE, Apt 1612, Atlanta, GA, 30308, United States, jyan40@gatech.edu, Turgay Ayer, Pinar Keskinocak, Aaron Caughey Down syndrome (DS) is a common type of chromosomal abnormality. Currently a one-size-fits-all type of risk-cutoff value of 1/270 is commonly used in DS screening to identify high-risk women and recommend an invasive confirmatory test, such as amniocentesis. In this study, we construct modeling frameworks to determine the optimal cutoffs from two practical perspectives of DS screening: one is the fairness across ages, and the other is the heterogeneity in women’s preferences about adverse pregnancy outcomes. We have shown the potential to improve health outcomes and patient satisfaction. Identification Of Optimal Partition For Semidefinite Optimization Ali Mohammad Nezhad, PhD Candidate, Lehigh University, 200 West Packer Ave, Bethlehem, PA, 18015, United States, alm413@lehigh.edu, Tamas Terlaky The concept of optimal partition was originally introduced for linear optimization and linear complementary problems and subsequently extended to semidefinite optimization. In this paper, under no assumption on strict complementarity, we formalize the optimal partition concept for semidefinite optimization. The magnitude of the eigenvalues belonging to each partition is quantified using a condition number and the degree of singularity of the problem. Wasserstein Distance And The Distributionally Robust Tsp Mehdi Behroozi, Assistant Professor, Northeastern University, 334 Snell Engineering Center, 360 Huntington Avenue, Boston, MA, 02115, United States, m.behroozi@neu.edu Recent research on the robust and stochastic Euclidean travelling salesman problem has seen many different approaches for describing the region of uncertainty, such as taking convex combinations of observed demand vectors or imposing constraints on the moments of the spatial demand distribution. In this paper, we consider a distributionally robust version of the Euclidean travelling salesman problem in which we compute the worst-case spatial distribution of demand against all distributions whose Wasserstein distance to an observed demand distribution is bounded from above. This constraint allows us to circumvent common overestimation that arises when other procedures are used. Reformulating The Disjunctive Cut Generating Linear Program Thiago Serra, Carnegie Mellon University, 5000 Forbes Ave, Pittsburgh, PA, 15213, United States, tserra@cmu.edu, Egon Balas, Francois Margot In lift-and-project, CGLP optima may yield dominated cuts due to distortions caused by normalization. This work proposes the Reverse Polar CGLP, a reformulation where normalization defines separability and the resulting cut minimizes the slack for a point in the disjunctive hull. Cuts derived from RP- CGLP optima define supporting hyperplanes of the disjunctive hull, hence never being strongly dominated. For a point in the interior of the disjunctive hull, the cutting plane is a combination of facets separating the fractional solution. We show equivalent CGLP formulations, explain the benefits of RP-CGLP, and present computational experiments where a larger gap can be closed after two rounds. Shahab Derhami, PhD Candidate, Auburn University, 3301 Shelby Center, Auburn, AL, 36849, United States, shahab.derhami@hotmail.comJeffrey Smith, Kevin Gue

181