Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

TB38

n TB34 North Bldg 223 10:30 - 11:15 Frontline Systems, Inc/ 11:15 - 12:00 Artelys Corp. Vendor Demo Session 1 - Tutorial: Preview of Analytic Solver V2019 for Windows, Macintosh, and Office 365 Daniel H. Fylstra, Frontline Systems, Inc., P.O. Box 4288, Incline Village, NV, 89450-4288, United States For many years, Frontline Systems has offered powerful data mining and machine learning, Monte Carlo simulation and risk analysis, and conventional and stochastic optimization in Analytic Solver« for Excel - on Windows. But use of Analytic Solver for Excel on Macintosh and the Web has involved challenges, due to differences within Excel. That’s changing for the better in Analytic Solver V2019, which has been rewritten to work with new Excel internal APIs from Microsoft. Attend this session to see a preview of Analytic Solver V2019: uniformly powerful and easier to use than ever - on Excel for Windows, Macintosh, and Office 365. 2 - Introducing the New API and Conic Solver in Artelys Knitro 11.0 Richard Waltz, Artelys, 150 N. Michigan Avenue, Suite 800, Chicago, IL, 60601, United States Artelys Knitro is the premier solver for nonlinear optimization problems. This software demonstration will highlight two key features in the new, major Knitro 11.0 release. First, we will demonstrate the new callable library API. This new API allows the user to build-up a model in pieces while providing special structures to Knitro. Second, we will introduce the new solver in Knitro 11.0 specially designed for models with cone constraints. Some benchmarking results will be provided. Chair: Lishuai Li, City University of Hong Kong, Kowloon, Hong Kong 1 - A Supervised Machine Learning Approach for Solving the Aircraft Recovery Problem Navid Rashedi, Dartmouth College, 30.5 W. Wheelock Street, 1, Hanover, NH, 03755, United States, Vikrant Vaze Commercial solvers have been used for solving aircraft recovery problems. However, obtaining high quality solutions within the limited time frames of online applications is challenging. We develop a data-driven approach to solve a broad class of aircraft recovery problems. This method can be applied to identify a near-optimal solution by using previously good solutions generated by decision- support tools, human experts, or a combination thereof. The performance of our method was tested by applying it to a set of real-world networks for a moderately large US carrier. The results showed that the obtained solutions were, on average, within a 5% optimality gap and were calculated within a few seconds. 2 - Characterization of Air Traffic Management Operations Based on Large-scale Flight Tracking Data Lishuai Li, City University of Hong Kong, Tat Chee Avenue, P6606, AC1, Kowloon, Hong Kong, Pan Ren Air Traffic Management (ATM) strategies and procedures vary by region. Comparing ATM in different regions could help us understand what works better and how to improve. However, few studies have done so due to lack of data. With Automatic Dependent Surveillance - Broadcast (ADS-B) adopted by many countries, it is possible for the first time to track and analyze aircraft movement data at global scale. I will present case studies to characterize actual ATM operations, i.e. network structure and dynamics, flow patterns, etc., via a data- driven approach. The result will enable us to understand current operations better, identify deficiencies, and provide recommendations for future improvement. n TB35 North Bldg 224A Data-Enabled Discovery and Applications in Air Transportation Sponsored: Aviation Applications Sponsored Session

n TB36 North Bldg 224B Joint Session Drones/Practice Curated: New Directions in Logistics Emerging Topic: Robotics, Drones and Autonomous Vehicles in Logistics Emerging Topic Session Chair: Dongdong Ge, Shanghai University of Finance and Economics 1 - An Efficient MILP Formulation for Real AGV Systems Dongdong Ge, Shanghai University of Finance and Economics, School of Information Management and Eng, 777 Guoding Road, Shanghai, 200433, China Running more than 100 robots in an AGV system is the main difficulty in most AGV warehouses. This project approaches this real time control problem by efficiently formulating its assignment part as an MILP problem, solving it efficiently, and combining it with intelligent partition system and vehicle routing algorithms. The implementation results in Beijing’s large warehouse in 2017 John Gunnar Carlsson, University of Southern California, 3750 McClintock Avenue, Los Angeles, CA, 90089, United States We determine the efficiency of a delivery system in which a truck acts as a “moving warehouse” for a collection of drones or ground-based robots. Using a theoretical analysis in the Euclidean plane, we derive a closed-form expression describing the time to completion in terms of the number of drones or robots as well as the relative speeds of the vehicles. n TB37 North Bldg 225A APS-Special Speaker Sponsored: Applied Probability Sponsored Session Chair: Harsha Honnappa, Purdue University, West Lafayette, IN, 47906, United States Co-Chair: Henry Lam, Columbia University, New York, NY, 10027, United States 1 - Data-integrated Stochastics: Models And Methods George Shanthikumar, Purdue University, West Lafayette, IN, 47906 This tutorial will review the current data integrated approaches for predictive and prescriptive analysis of stochastic systems. In particularly we will review: 1) approaches such as Multi-Armed Bandit, Regularization in Sample Average Approximation and Data Driven Robust Optimization for generating prescriptive solutions to stochastic systems, and 2) some of the Machine Learning approaches used for predictive analysis of stochastic systems. We will then provide a framework for data integrated methodology for predictive and prescriptive analytics for stochastic systems. Specific attention will be paid to overcoming structural and statistical errors. n TB38 North Bldg 225B Joint Session APS/Practice Curated:Data Driven OR Applications Sponsored: Applied Probability Sponsored Session Chair: Flora Spieksma, Leiden University, Leiden, 2311KL, Netherlands Co-Chair: Michael N. Katehakis, Rutgers University, Newark, NJ, 07010, United States 1 - Non-parametric Up-and-down Experimentation Revisited Flora Spieksma, Associate Professor, Leiden University, Niels Bohrweg 1, Leiden, 233CA, Netherlands, Michael N. Katehakis, Sheldon M. Ross In many applications, it is of interest to estimate a value, the æquantile’, of a scalar control variable x that will cause a certain ratio a of successes, or failures, in a binary output variable Y = Y (x). We present a convergent algorithm for estimating the quantile x* and we discuss applications in inventory control under unknown demand distribution, and drug dosage determination. showed a great success for ‘double eleven’ shopping rush. 2 - Last Mile Delivery using Trucks, Drones, and Ground-based Robots

287