Informs Annual Meeting 2017

SD82

INFORMS Houston – 2017

SD82

different techniques. Numerical results will be presented. 2 - Stochastic Dual Dynamic Integer Programming

382B Online Algorithms and Applications Sponsored: Optimization, Optimization Under Uncertainty Sponsored Session Chair: Shashank Goyal, University of Minnesota, Minneapolis, MN, 55414, United States, goyal030@umn.edu 1 - Time-indexed Relaxations for the Online Bipartite Matching Problem Alfredo Torrico, Georgia Institute of Technology, 201 16th Street NW, Apartment 1, Atlanta, GA, 30363, United States, atorrico3@gatech.edu, Alejandro Toriello We study an i.i.d. model of the classical online bipartite matching problem, in which one side of the bipartition is fixed and known in advance, while vertices from the other side appear sequentially as i.i.d. realizations of a uniform distribution. Each one of these arrivals must be matched or discarded. We study several relaxations in a time-indexed space. We derive valid inequalities for the polyhedron of achievable matching probabilities, and discuss when they are facet- defining. In contrast to previous works, we show that our inequalities either theoretically dominate the constraints presented before, or provide empirically superior bounds. 2 - Online Covering with Norm Objectives and Applications Xiangkun Shen, University of Michigan, Ann Arbor, MI, 48109, United States, xkshen@umich.edu, Viswanath Nagarajan We consider fractional online covering problems with lq-norm objectives. The covering constraints arrive online over time. We provide an online log- competitive algorithm. This is based on the online primal-dual framework where we use the dual program. Our result expands the class of convex objectives that admit good online algorithms: prior results required a monotonicity condition on the objective which is not satisfied here. This result is nearly tight even for the linear special case. As direct applications, we obtain (i) improved online algorithms for non-uniform buy-at-bulk network design and (ii) the first online algorithm for throughput maximization under lq-norm edge capacities. 3 - Online Algorithms for Rental Requests Shashank Goyal, University of Minnesota, 1071 17th Ave SE, Minneapolis, MN, 55414, United States, goyal030@umn.edu, Diwakar Gupta Web platforms such as Airbnb and Craigslist provide individuals the ability to post availability of a resource they own and rent it to different customers for short periods of time. A customer can place a request with the owner, indicating the desired rental start time and duration. The owner must then decide whether to accept or decline the request. A key trade-off that the owner confronts is the reward he earns from the current request and the potential loss of reward from a higher-value future requests that may clash with the current one. We propose online algorithms for the problem that have provable worst-case performance and also present bounds on the best performance that any algorithm can achieve. 4 - Online Optimization of Google Ad-click Auctions Donghun Lee, Princeton University, 9 Lawrence Drive, Apt 202, Princeton, NJ, 08540, United States, donghunl@princeton.edu, Warren B. Powell We present online optimization framework to maximize the return of advertisement campaign through repeated real-time bidding auctions under budget constraints. The framework learns the properties of the auction environment online, and makes on-the-fly optimization with optimal learning algorithms. We test the performance of the framework by testing on real-time bidding auction simulator based on Google Adwords platform history of placing hotel advertisements on the web. 382C Stochastic Integer Programming Sponsored: Optimization, Optimization Under Uncertainty Sponsored Session Chair: Jim R Luedtke, University of Wisconsin-Madison, Madison, WI, 53706, United States, jim.luedtke@wisc.edu 1 - Dual Decision Rules for Stochastic Integer Programming Jim R.Luedtke, University of Wisconsin-Madison, 3236 Mechanical Engineering, 1513 University Ave, Madison, WI, 53706, United States, jim.luedtke@wisc.edu, Merve Bodur We consider decision rules for multi-stage stochastic integer programming problems. We investigate techniques for using these decision rules to obtain bounds on the optimal solution, and compare the strength of the relaxation from SD83

Shabbir Ahmed, Georgia Institute of Technology, 765 Ferst Drive, Atlanta, GA, 30332, United States, sahmed@isye.gatech.edu We extend the Stochastic Dual Dynamic Programming (SDDP) algorithm to solve multistage stochastic integer programs with binary state variables. Various case studies in energy systems is discussed. 3 - Two-stage Stochastic MIPS with Structured Non-linear Programs Yingqiu Zhang, Graduate Student, Virginia Tech., Blacksburg, VA, 24060, United States, yqzhang7@vt.edu, Manish Bansal We present our recent developments for solving two stage stochastic mixed integer programs with structured mixed integer non-linear programs in the second stage. We utilize globally valid parametric cuts within Benders’ decomposition to tighten the second stage programs, and provide results of our computational experiments.

Monday, 8:00 - 9:30AM

MA01

310A Strategic Decisions Sponsored: Decision Analysis Sponsored Session

Chair: Dharma Kwon, University of Illinois at U-C, University of Illinois at U-C, Champaign, IL, 61820, United States, dhkwon@illinois.edu Co-Chair: Youngsoo Kim, University of Illinois at Urbana-Champaign, Champaign, IL, 61820, United States, ykim180@illinois.edu 1 - It is Time to Get Some Rest Manel Baucells, University of Virginia, Darden School of Business, 100 Darden Boulevard, Charlottesville, VA, 22903, United States, mbaucells@gmail.com, Lin Zhao We introduce the fatigue disutility model in continuous-time. It captures in a precise way the notion that fatigue accumulates with effort, decays with rest, and increases marginal instant disutility. While prescriptive, the model rationalizes several behavioral anomalies. Under very general conditions the most efficient temporal profile of effort exhibits a high-low-high pattern. We examine the optimal management of fatigue and productivity under various scenarios. We bring the model to the data and provide a calibration that predicts well the speed profiles observed in swimming competitions. 2 - Product Management and Innovation in a Supply Chain Junghee Lee, University of California, San Diego, 9500 Gilman How can we achieve broader market coverage for innovative products, i.e., inclusive innovation? Grounded in industrial practice, we show that deliberately choosing the contract leader and the investor in a multi-tiered supply chain can have a significant impact on market coverage. We discuss leadership handovers along the product life cycle. 3 - Audit and Remediation Strategies in the Presence of Evasion Capabilities Shouqiang Wang, The University of Texas at Dallas, Naveen Jindal School of Management, 800 W. Campbell Rd, Richardson, TX, 75080, United States, Shouqiang.Wang@utdallas.edu, Francis E. De Vericourt, Peng Sun Companies often have means and incentive to conceal a negative issue, which, if not addressed promptly, is likely to result in detrimental consequences. Such evasive actions render audits no longer effective in revealing such an issue. We Examine how to uncover and remedy it in the presence of evasive actions. 4 - Strategic Investment in Shared Suppliers with Quality Deterioration Youngsoo Kim, University of Illinois at U-C, 1206 South Sixth Street, 350 Wohlers Hall, Champaign, IL, 61820, United States, ykim180@illinois.edu, Dharma Kwon, Anupam Agrawal Motivated by manufacturers’ quality investment in shared suppliers, we study an investment game with spillover and uncertainty. Each manufacturer repeatedly decides either to invest in the declining quality of the supplier or to wait for the other to do so. We find that a unique mixed strategy equilibrium exists if and only if asymmetry in investment costs is not too large. Moreover, a mixed strategy equilibrium disappears if investment opportunities are not repeated. We then consider Nash bargaining as a resolution for the potential inefficiency from a mixed strategy equilibrium. Using the field-collected data, we demonstrate the efficacy of bargaining to avoid a mixed strategy equilibrium. Drive, #0553, La Jolla, CA, 92093-0553, United States, junghee.lee@rady.ucsd.edu, Hyoduk Shin, Vish Krishnan, Oleksiy Viktorovych Mnyshenko

130