2015 Informs Annual Meeting

WD13

INFORMS Philadelphia – 2015

WD12 12-Franklin 2, Marriott Optimization Stochastic IV Contributed Session Chair: Jeremy Castaing, University of Michigan, IOE 1205 Beal Ave., Ann Arbor, MI, United States of America, jctg@umich.edu 1 - Application of Two-stage Risk-averse Optimization to Real-time Interday Portfolio Management Sitki Gulten, Assistant Professor, Stockton University, 101 Vera King Farris Drive, Galloway, NJ, United States of America, sitki.gulten@stockton.edu, Andrzej Ruszczynski We describe a study of application of risk-averse optimization techniques to daily portfolio management. First, we develop clustering methods for scenario tree construction. Then, we construct a two-stage stochastic programming problem with conditional measures of risk, which is used to re-balance the portfolio on a rolling horizon basis, with transaction costs included. Finally, we present an extensive simulation study on both intraday and interday real-world data of the methodology. 2 - Stochastic Mixed Integer and Gradient Search Methods for Constrained Problems Larry Fenn, Hunter College, CUNY, 695 Park Avenue, New York, NY, 10021, United States of America, larry.fenn@gmail.com, Felisa Vazquez-abad An ordinal variable b denotes resources. For each b, there is a control variable u that yields a convex cost C(b,u). We seek minimal cost satisfying a constraint of the form P(b,u) < a. Both functions are steady state averages of a stationary complex process: function evaluations are very costly. We combine Fibonacci search in b with gradient search in u to find the optimal solution and propose sequential sampling for minimal computation. We analyze convergence and discuss parallel computation. 3 - Optimally Scheduling Satellite Communications under Uncertainty Jeremy Castaing, University of Michigan, IOE 1205 Beal Ave., Ann Arbor, MI, United States of America, jctg@umich.edu, Amy Cohn, James Cutler We consider the problem of scheduling and managing the download of data from collecting satellites to receiving ground stations under uncertainty of their availability. We design models and heuristics to compute download schedules over the planning horizon while addressing dynamics of collecting, storing, using, and spilling both data and energy. WD13 13-Franklin 3, Marriott Stochastic Programming Sponsor: Optimization/Optimization Under Uncertainty Sponsored Session Chair: Guzin Bayraksan, Associate Professor, The Ohio State University, Integrated Systems Engineering, Columbus, OH, 43209, United States of America, bayraksan.1@osu.edu 1 - Regularized Decomposition of High–dimensional Multistage Stochastic Programs with Markov Uncertainty Tsvetan Asamov, Princeton University, Sherrerd Hall, Charlton Street, Princeton, NJ, 08544, United States of America, tasamov@princeton.edu, Warren Powell We develop a quadratic regularization approach for the solution of high–dimensional multistage stochastic optimization problems. The resulting algorithms are shown to converge to an optimal policy after a finite number of iterations. Computational experiments are conducted using the setting of optimizing energy storage over a large transmission grid. The numerical results indicate that the proposed methods exhibit significantly faster convergence than their classical counterparts. 2 - Multistage Stochastic Optimization with Application in Energy Storage Control Harsha Gangammanavar, University of Southern California, 3715, McClintock Avenue, GER 240, Los Angeles, CA, 90089, United States of America, gangamma@usc.edu, Suvrajeet Sen In this talk we will present Time-staged Stochastic Decomposition, a sequential sampling algorithm which is applicable to multistage stochastic linear programs, in general, and control of distributed storage devices, in particular. The method is focused on stage independent uncertainty and its special cases, including autoregressive processes. We will present convergence properties of this algorithm, and computational results comparing it with algorithms like approximate dynamic programming.

4 - Managing Cross-border E-commerce: A Multichannel Integration Approach Shikui Wu, Assistant Professor, Ryerson University, 350 Victoria St., TRS 2-015, Toronto, Canada, shikui.wu@ryerson.ca, Yuanyuan Wu This study adopts a multichannel integration approach to analyze and optimize online and offline business strategies and operations for cross-border e-commerce, including: pricing and inventory control; sales, deliveries and returns; and, online and in-store customer service. It helps cross-border businesses in adapting their strategies to market environment, streamlining their business processes, reducing operational costs, mitigating market uncertainty, and improving customer satisfaction.

WD11 11-Franklin 1, Marriott Optimization Integer Programming III Contributed Session

Chair: Carlos Eduardo De Andrade, Senior Inventive Scientist, AT&T Labs Research, 200 Laurel Avenue South, A5-1E33, Middletwon, NJ, 07748, United States of America, cea@research.att.com 1 - Cardinality Constrained Portfolio Optimization via Mixed Integer Linear Programming Nasim Dehghan Hardoroudi, Aalto University School of Business, Runeberginkatu 22-24, Helsinki, N/, 00100, Finland, nasim.dehghan@gmail.com, Abolfazl Keshvari Controlling the number of active assets (cardinality of the portfolio) in a mean- variance portfolio problem is practically important but computationally difficult. Such task is commonly presented as a mixed integer quadratic programming (MIQP) problem. We propose a novel approach to reformulate such problem as a MILP problem. Our proposed formulation can be solved by standard MILP solvers much faster than the MIQP problem. 2 - On Linear Conic Relaxation of Discrete Quadratic Programming Tiantian Nie, North Carolina State University, Department of Industrial and Systems, Engineering, Raleigh, NC, 27695, United States of America, tnie@ncsu.edu, Shu-cherng Fang, Qi An Discrete quadratic programming problems (DQP) appear widely in real-life applications, but they are hard to solve. We proposed an RLT-based linear conic relaxation of DQP. When the proposed relaxation problem has an optimal solution with rank one or two, optimal solutions to the original DQP problem can be explicitly generated. Numerical results indicate that the proposed relaxation with a primal ADM procedure is capable of efficiently providing high-quality lower bounds for DQP. 3 - Location Problem of Electric Vehicles’ Charge Stations under Uncertainty Rui Chen, IE Department of Tsinghua University, ShunDe Building in Tsinghua University, Beijing, China, chenruiest@163.com The location of charging stations spreads adoption of electric vehicles by consumer and it simultaneously affects the cost and revenue. This paper presents a location problem for electric vehicles’ charge stations under fluctuation of electricity price. An integer programming model is proposed to solve this problem with uncertainty of demand. Then a hybrid algorithm is applied to and numerically verified by an actual example. 4 - A Learning Framework for Feasibility Pump Carlos Eduardo De Andrade, Senior Inventive Scientist, AT&T Labs Research, 200 Laurel Avenue South, A5-1E33, Middletwon, NJ, 07748, United States of America, cea@research.att.com, Shabbir Ahmed, Yufen Shao, George L. Nemhauser We present a framework for finding feasible solutions for mixed integer programs. We use the feasibility pump heuristic (FP) coupled with a learning framework which is able to build a pool of projections and combine them using information of previous projections. Preliminary results shows that this approach is able to find feasible solutions for instances where the original FP fails.

457