2015 Informs Annual Meeting

TC75

INFORMS Philadelphia – 2015

TC76 76-Room 204C, CC Advances in Stochastic Simulation

3 - Discussant’s Presentation Yu Ding, Professor, Texas A&M University, ETB 4016, MS 3131, College Station, TX, United States of America, yuding@iemail.tamu.edu As a discussant in this Technometrics special issue session on system informatics, I will present my understanding of strengths and weaknesses of the two papers selected by Technometrics editor for this session. I will also discuss other related research problems on the similar topics. TC75 75-Room 204B, CC IBM Research Best Student Paper Award III Sponsor: Service Science Sponsored Session Chair: Ming-Hui Huang, National Taiwan University, Taiwan - ROC, huangmh@ntu.edu.tw 1 - Best Student Paper Competitive Presentation Ming-Hui Huang, National Taiwan University, Taiwan - ROC, huangmh@ntu.edu.tw Finalists of the IBM Research Best Student Paper Award present their research findings in front of a panel of judges. The judging panel will decide the order of winners, which will be announced during the business meeting of the Service Science Section at the Annual Conference. 2 - Efficient Information Heterogeneity in a Queue Yang Li,Rotman School of Management, University of Toronto, 105 St. George Street, Toronto ON M5S3E6, Canada, Yang.Li10@Rotman.Utoronto.Ca, Ming Hu How would the growing prevalence of real-time delay information affect a service system? We consider an M/M/1 queueing system in which only a fraction of customers are informed about real-time delay. Surprisingly, we find that system throughput and social welfare can be unimodal in the fraction of informed customers. 3 - Scheduling and Pricing Services for Online Electric Vehicle Charging Mark Nejad,Assistant Professor, University of Oklahoma, Industrial and Systems Engineering, Norman OK, United States of America, mark.nejad@ou.edu , Ratna Babu Chinnam, Daniel Grosu, Lena Mashayekhy We design mechanisms for EV charging services in online settings. We prove that our proposed mechanisms are incentive compatible, that is, truthful reporting of price and the amount of charging is a dominant strategy for self-interested EV drivers. Our preemption-aware charging mechanisms allow providers to manage fluctuations in renewable energy production. 4 - Scheduling with Testing Thomas Magnanti,Institute Professor, MIT, 77 Massachusetts Avenue, 32-D784, Cambridge MA 02139, United States of America, magnanti@mit.edu, Retsef Levi, Yaron Shaposhnik We study a new class of scheduling problems that captures a common tradeoff between using resources for processing jobs, and investing resources to ‘test’ jobs and learn more about their uncertain attributes. This can inform future decisions, but also delay service. We derive intuitive structural properties of the optimal policies, and use a new cost-accounting scheme to devise a surprisingly low dimensional dynamic programming formulation, which ultimately leads to an FPTAS. 5 -Trading Time in a Congested Environment Luyi Yang,Doctoral Student, University of Chicago Booth School of Business, Chicago, IL, United States of America, luyi.yang@chicagobooth.edu, Laurens Debo, Varun Gupta We propose a time-trading mechanism, mediated by a revenue maximizing broker, in which customers privately informed about their waiting costs mutually agree on the ordering in a queue via trading positions. To that end, we show that the broker can implement an auction with a trade-participation fee and two trade restriction prices on customer bids. Under the optimal auction, there is partial pooling in the bidding strategies and therefore customers are not strictly prioritized.

Sponsor: Simulation Sponsored Session Chair: Henry Lam, University of Michigan, 1205 Beal Ave., Ann Arbor, MI, United States of America 1 - Risk Assessment for Input Uncertainty Helin Zhu, School of Industrial and Systems Engineering, Georgia Institute of Technology, 755 Ferst Drive NW, Atlanta, GA, 30332, United States of America, hzhu67@gatech.edu, Enlu Zhou When simulating a complex stochastic system, the behavior of the output response depends on the input parameter estimated from finite real-world data, and the finiteness of data brings input uncertainty to the output response. Risk assessment for input certainty, which quantifies the extreme behavior of the mean output response, is extremely important. In the present paper, we introduce the risk measures for input uncertainty and study the corresponding estimators. 2 - Projected Directional Derivatives for High Dimensional Gradient Estimation Raghu Pasupathy, Associate Professor, Department of Statistics, Purdue University, 250 N University Street, West Lafayette, IN, 47907, United States of America, pasupath@purdue.edu, Boqian Zhang We present a method to estimate gradients in high dimensions by projecting randomly generated directional derivatives onto the various axes. We discuss theoretical properties and sampling measures that minimize the resulting estimator’s error norm. The method appears particularly relevant in high dimensions since only two observations are needed for a complete gradient estimator. 3 - Perfect Sampling of GI/GI/C Queues Yanan Pei, Columbia University, 500 W. 120th St, Mudd 313, We introduce the first class of perfect sampling algorithms for the steady-state distribution of multi-server queues with general inter-arrival time and service time distributions. Our algorithm is built on the classical dominated coupling from the past protocol using a coupled multi-server vacation system as the upper bound process. The algorithm has finite expected termination time with mild moment assumptions on the inter-arrival time and service time distributions. 4 - Rare Event Simulation in the Neighborhood of a Rest Point Konstantinos Spiliopoulos, Assistant Professor, Boston University, Department of Mathematics and Statistics, 111 Cummington Mall, Boston, MA, 02215, United States of America, kspiliop@math.bu.edu We construct efficient importance sampling Monte Carlo schemes for finite time exit probabilities in the presence of rest points. The main novelty of the work is the inclusion of rest points in the domain of interest. We motivate the construction of schemes that perform well both asymptotically and non- asymptotically. We concentrate on the regime where the noise is small and the time horizon is large. Examples and simulation results are provided. Joint work with Paul Dupuis and Xiang Zhou. New York, NY, 10027, United States of America, yp2342@columbia.edu, Jose Blanchet, Jing Dong

TC77 77-Room 300, CC Logistics II Contributed Session

Chair: Fateme Fotuhiardakani, Data Scientist, TMW Systems, 6085 Parkland Blvd, Mayfield Heights, OH, 44124, United States of America, fateme.fotuhi@gmail.com 1 - Using Heuristics to Solve the Container Loading Problem Focusing on Priority Levels and Utilization Crystal Wilson, Clemson University, 6 Natalie Ct., Greer, SC, 29651, United States of America, crysta3@clemson.edu, Mary Beth Kurz Just-in-time manufacturers need the parts to arrive to the facility by a scheduled time to keep the assembly line moving smoothly. How small containers, such as parts, are loaded onto a larger container is a special type of packing problem. This research will focus on creating a heuristic that creates loading patterns that balances priority levels, while also maximizing the utilization of the container with respect to the weight and cube.

344