Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

WD16

Experiments on problem instances with up to 6000 variables demonstrate the effectiveness of the proposed algorithm. 2 - Genetic Algorithm for Planar Capacitated Single Allocation Hub Location Problem Derya Ipek Eroglu, Middle East Technical University, Ankara, 06450, Turkey, Duygu Pamukcu, Cem Iyigun In this study, Genetic Algorithm (GA) has been implemented to Planar Capacitated Single Allocation p-Hub Location Problem. In this problem, the objective is to locate the hubs and assign the supply and/or demand points to hubs so as to minimize the total of distribution and transfer costs. If Euclidean Distance is used as a distance measure calculating the cost, the problem becomes nonlinear. Therefore, GA has been implemented to the problem. Computational studies have been performed regarding solution quality and runtime, and solutions are compared with these of Discrete Capacitated Single Allocation p- Hub Location Problem. 3 - A Study on Two-stage Genetic Algorithm for the Mixed-integer Linear Programming with Non-linear Objective Functions Kiseok Sung, Gangneung-Wonju National University, 150 NamWon-ro, Wonju, 26403, Korea, Republic of We propose a genetic algorithm for MILP with nonlinear objective function. This algorithm consists of two stages of a hierarchical evolutionary process. At higher levels, integer variables evolve and linear variables evolve at lower levels, where we use a method that maintains the feasibility of a linear variable according to a linear equation constrained by a given integer chromosome. This method uses null-space projection and bound shift techniques. I have been coding the proposed method in MATLAB and tested the sample problems. 4 - Revenue Management for Dual Channel Retailer Using Randomized Decomposition Based Evolutionary Algorithm Shrinath Dakare, Indian Institute of Technology-Kharagpur, Meghnad Saha Hall of Residence, Kharagpur, 721302, India, Vishal Gupta, Manoj Kumar Tiwari We consider integrated price optimization, inventory replenishment and demand fulfillment problem for a dual channel retailer who has adopted cross channel fulfillment strategy. Demand function is modeled using MNL model with price dependent attraction function. The problem so formulated is non-linear and non- convex that we solved using Randomized Decomposition based Evolutionary Algorithm. We compare the results obtained with that of recently proposed Randomized Decomposition based Neighborhood Search. We also demonstrate that price optimization problem could efficiently be solved using intelligent search heuristics, technique which has largely been ignored for such problems. 5 - A Heuristic Approach to Passenger-oriented High-speed Train Schedule Model Jiemin Xie, The University of Hong Kong, Hong Kong, China, Shuguang Zhan, S.C. Wong, S.M. Lo A better railway timetable for high-speed railway, HSR, can provide a more convenient service to passengers, which may attract more consumers transferring from domestic air flights and cars. In this study, a schedule model, which minimizes the total path cost of passengers and is subjected to both standard train schedule and passenger path choice constraints, is proposed to generate conflict- free and convenient schedule plan. A heuristic HSR scheduling algorithm is proposed to solve our model. This heuristic algorithm is based on the decomposition approach and the iterative approach. The South China HSR network is used to test the optimality, efficiency, and applicability. n WD19 North Bldg 128B Joint Session RMP/Practice Curated: New Applications of Revenue Management and Pricing Sponsored: Revenue Management & Pricing Sponsored Session Chair: Joline Uichanco, University of Michigan, Ross School of Business, Ann Arbor, MI, 48109-1234, United States 1 - Price Markdowns to Induce Customers to Opt Out of Free Returns Sajjad Najafi, Ross School of Business, University of Michigan, 701 Tappan Ave, Ann arbor, MI, 48109, United States, Izak Duenyas In this paper, we consider a firm offering a product to a set of consumers who can return it over a grace period of time. The firm aims at finding the optimal return policy to improve its total expected revenue.

n WD16 North Bldg 127B Economic Modeling Contributed Session Chair: Fei Qin, Shippensburg University, John L. Grove College of Business, 1871 Old Main Drive, Shippensburg, PA, 17257, United States 1 - A Spatial Model of Ride Hailing Cemil Selcuk, Cardiff University, Aberconway Building, Colu, Cardiff, CF10 3EU, United Kingdom We build a spatial model to study the selection and performance of price and commission policies for a ride-sharing platform. We show that a flexible commission policy is more effective in matching demand and supply than a flexible price policy, especially if the platform does not have many cars at its disposal. Without flexible commissions, the platform resorts to price hikes to prevent excess demand; however such interventions not only distort the unconstrained demand and lower profits, but also do a poor job in spreading drivers across the city. In contrast, with a flexible commission policy the platform can spread the cars around without distorting the demand and running into any bottlenecks. 2 - A Marketplace for Data: An Algorithmic Solution Anish Agarwal, Machusetts Institute of Tehnology, Cambridge, MA, United States, Munther Dahleh, Tuhin Sarkar We aim to create a data marketplace; a robust matching mechanism to efficiently buy and sell data while optimizing social welfare and maximizing revenue. Data is a unique asset: it is replicated at zero marginal cost; value is inherently combinatorial i.e. value of a dataset depends on what other data is available; value for data vary widely between firms; authenticity of data cannot be verified a priori. Hence, setting prices for a collection of datasets with correlated signals is non-trivial. Our proposed marketplace provides a complete algorithmic solution combining concepts from algorithmic market design and online optimization. 3 - Strategic Inventories in a Supply Chain with Downstream Cournot Duopoly Xiaowei Hu, PhD Student, University of Wisconsin-Milwaukee, Milwaukee, WI, 53212, United States We present a game-theoretic supply chain model, in which a manufacturer supplies products to a pair of identical Cournot duopolistic retailers. We show that the strategic inventory carried out by the retailers under dynamic contract is Pareto-dominating for the manufacturer, retailers, consumers, the channel, and the society as well. We also find that retailer’s strategic inventory can be eliminated when the manufacturer commits wholesale contract or inventory holding cost is too high. In comparing the cases with and without downstream competition, we also show that the downstream Cournot duopoly undermines the profits for the retailers, but benefits all others. 4 - Engineering Economic Valuation of Blockchain Technology for Petroleum Industry Supply Chain under Uncertainty K. Jo Min, Iowa State University, 3004 Black Engineering, Department of Industrial &, Ames, IA, 50011-2164, United States, John Jackman, Farshad Niayeshpour Much interests in blockchain technology applicable to petroleum industry supply chain notwithstanding, there has been little analytical investigation of the critical attributes of the blockchain technology such as the value of immutability, irrefutability, traceability, scalability, and transparency. In this paper, we aim to show how such attributes can be valued, in the context of petroleum industry supply chain, from a perspective of stochastic optimal control. In particular, we will show what the underlying assets may be and how strike prices can be found via real options approach. Illustrative numerical examples based on the fluctuation of the petroleum product prices will be presented. Metaheuristics Contributed Session Chair: Jiemin Xie, University of Hong Kong, 91B Pokfulam Road, Pokfulam, Hong Kong, 999077, China 1 - An Effective Memetic Algorithm for the Boolean Quadratic Programming Problem with Generalized Upper Bound Constraints Daiqiang Yin, Northwestern Polytechnical University, Xi’an, China, Yang Wang, Abraham Punnen We propose an effective memetic algorithm for solving the NP-hard boolean quadratic programming problem with generalized upper bound constraints. The main ingredients of the proposed algorithm include a reverse learning based population initialization, a uniform combined with greedy rule guided crossover operator, an adaptive tabu search method and a rank based population updating. n WD18 North Bldg 128A

500