Informs Annual Meeting Phoenix 2018

TE10

INFORMS Phoenix – 2018

2 - A First Order Method for Stochastic Variational Inequalities on Semidefinite Matrix Spaces Nahidsadat Majlesinasab, Oklahoma State University, Stillwater, OK, United States, Farzad Yousefian, Mohammad Javad Feizollahi Motivated by multiuser noncooperative Nash games in stochastic regimes, we consider stochastic variational inequality problems (SVI). While the literature of variational inequality (VI) has focused on vector spaces, there is not enough guidance on solution methods for addressing SVIs on semidefinite matrix spaces. Motivated by this gap, we develop a stochastic mirror descent method. Our contribution is three-fold: (i) employing averaging techniques, we show the generated iterate converges to a weak solution of the SVI; (ii) we derive a convergence rate in terms of the expected value of a gap function; (iii) we implement the method for solving a MIMO throughput maximization problem. 3 - An Iterative Regularized Mirror Descent Method for Illposed Nondifferentiable Stochastic Optimization Mostafa Amini, Oklahoma State University, Stillwater, OK, United States, Farzad Yousefian A wide range of big data applications result in optimization problems that are often illposed. To address such problems, we consider a model, where the goal is to find an optimal solution that attains the minimum value of a regularizer. We assume the objective function of the main problem and the regularizer are nondifferentiable convex functions, and the main objective is stochastic. We develop an iterative regularized stochastic mirror descent method. We establish the convergence of the iterate generated by the algorithm to the desired optimal solution in both an almost sure and a mean sense, and derive a convergence rate of optimality with respect to the main objective. 4 - A First Order Method for High Dimensional Illposed Problems Harshal Kaushik, Oklahoma State University, 322 Engineering North, Ind, Stillwater, OK, 74078, United States, Farzad Yousefian We consider high dimensional ill-posed convex optimization problems. To address ill-posedness, we consider minimizing a regularizer over the optimal solution set of a convex optimization problem. We develop a randomized block coordinate iterative regularized gradient method where at each iteration, the stepsize and regularization parameter are updated. We show that the generated sequence converges to the solution of the bilevel problem, and we derive a convergence rate of the order d/k^0.5 on the feasibility gap where k and d denote the iteration number and number of blocks, respectively. The performance of the algorithm for solving linear inverse problems in image deblurring is presented. n TE09 North Bldg 124B Managing Technology Products across Generations Sponsored: Manufacturing & Service Oper Mgmt Sponsored Session Chair: Charles X. Wang, State University of New York-Buffalo, Buffalo, NY, 14260, United States Co-Chair: Aditya Vedantam, SUNY Buffalo, Williamsville, NY, 14221, United States 1 - Remanufacturing Competition Dongmei Xue, Zhejiang University, Hangzhou, China The competition between firms is growing fierce,especially for the low-end firms. As the profit margin of low-end market decreasing, we provide remanufacturing as a new competition strategy for low-end firms to enter the high-end market. The goal of this paper is to find out how the optimal competition strategy is influenced by the cost structure and customers’ valuation. If high-end firm’s strategy is given, we show the optimal strategy space is divided into four pieces.Remanufacturing can be a profitable strategy when the cost is lower.Moreover, when high-end firm react to low-end firms’ strategy, the optimal strategy space shrink. 2 - Framework for Reducing Technical Debt Vera Tilson, University of Rochester, 3-343 Carol Simon Hall, W. E. Simon Graduate School of Business, Rochester, NY, 14627, United States, David Tilson We discuss a framework for reducing technical debt: retiring, outsourcing applications and starting to work on integrated data model to reduce coupling and moving toward “to-be” vision for the long-term architecture. 3 - Supplier Contracting for Reuse at a Third Party Remanufacturer Aditya Vedantam, State University of New York at Buffalo, 13 Via Pinto Drive, Buffalo, NY, 14221, United States, Ananth V. Iyer Business users of electronic products often contract with third parties (3PRs) in the secondary market to dispose their equipment at end-of-use. 3PRs provide services such as data security, value recovery and environmental reports to generators. Motivated by an extensive dataset provided by a 3PR, we generate

insights on contracts and drivers of disposition activity for IT equipment in the secondary market. We show how supplier contracts and reuse rates depend on planned replacement activities. Business users can encourage reuse by early planned replacement and modifying contracts. We discuss the environmental impact of planned vs. unplanned equipment replacement. 4 - Manufacturer’s Mental Accounting and Payment Schemes in Returns Policy Jun Ru, College of Business Administration, Cal Poly-Pomona, Pomona, CA, United States Returns policies have been used between the manufacturer and retailer in the distribution of short life-cycle products with uncertain demand. Previous research has shown that returns policies can mitigate the retailer’s risk of excess inventory and improve channel performance. This research extends our understanding of returns policies by adopting the concept of mental accounting to describe the manufacturer’s behavioral decisions under returns policies. We also investigate two alternative payment schemes (i.e., manufacturer financing and retailer financing) that help mitigate the manufacturer’s mental accounting effect in returns policies and improve channel performance 5 - Selling A Technology Product with the Trade-in Program and Used Product Market Charles X. Wang, State University of New York-Buffalo, Jacobs Management Center, Room 326M, Buffalo, NY, 14260, United States, Imsu Park, Mingcheng Wei We investigate the manufacturer’s optimal strategy for selling a technology product with the trade-in program and used product market. We characterize the interaction between the equilibrium price of the used product market and the trade-in program offered by the manufacturer in the primary market. n TE10 North Bldg 125A Innovative Retail Operations Sponsored: Manufacturing & Service Oper Mgmt Sponsored Session Chair: Hao-Wei Chen, University of Toledo, Toledo, OH, 43615, United States 1 - Cost Auditing in Profit Sharing Contracts Xiaomin Du, Tsinghua University, Beijing, China, Wanshan Zhu, Zhengping Wu This study is motivated by profit sharing contracts observed in a supplier-retailer channel, where knowledge of the supplier’s true production cost is the key to successful contract implementation. However, production cost is often the supplier’s private information, unknown to the retailer. We investigate the role of cost auditing mechanisms in such a setting. While cost auditing enables the retailer to protect his interests, interestingly, we find that the supplier may also benefit from it. We also examine supply chain performance under such mechanisms. 2 - Optimal Information Provision for Undifferentiated Products Huaqing Wang, Emporia State University, 18923 SW 7th Street, Pembroke Pines, FL, 33029, United States, Haresh B. Gurnani, Raphael Boleslavsky In this paper, we examine the joint interaction of information provision and pricing decisions by two competitive firms when a buyer is uncertain about product valuations. Firms generate product differentiation by allowing consumers to learn about valuations or prevent them from doing so. In this research, we characterize equilibrium prices and its interaction with information policies. 3 - Modeling Retail Store and Online Channel Allocation Decisions for Multiple Items Roshanak Mohammadivojdan, University of Florida, 303 Weil Hall, P.O. Box 116595, Gainesville, FL, 32611, United States, Joseph Geunes We consider a multi-product retailer who sells items via both a retail store and an online channel. Retail store shelf space limitations require judicious selection of items that will occupy this valuable space. Moreover, the high relative cost of this space, combined with the availability of a lower cost (in terms of storage-related costs) online channel, force a retailer to determine which items to offer in the store, online, or not at all. The goal of this work is to propose a stylized mathematical program to simultaneously model product assortment and channel allocation decisions, in order to gain insights on key decision drivers and characteristics of optimal decisions.

383