2016 INFORMS Annual Meeting Program

TC78

INFORMS Nashville – 2016

2 - Mixed-integer Programming For Detecting Cycles In Non-reversible Markov Processes Ambros Gleixner, Zuse Institute Berlin, Takustr. 7, Berlin, 14195, Germany, gleixner@zib.de, Isabel Beckenbach, Leon Eifler, Konstantin Fackeldey, Andreas Grever, Marcus Weber, Jakob Witzig Spectral clustering methods are used successfully to identify so-called metastable sets or dominant cycles of Markov processes. They analyze the spectrum of matrices based on the transition probabilities between states. However, current spectral methods have limited applicability for many biological and catalytic processes with a systematic, but slow cyclic behavior on the timescale of the simulation. We use mixed-integer programming to develop a new method that is able to detect the existence and measure the strength of such cyclic processes. 3 - Enhanced Equations For Linearizing Discrete Product For Generalized Programming Problems Qi An, North Carolina State University, 2119A Gorman Street, Raleigh, NC, 27606, United States, nival.ann@gmail.com, Tiantian Nie, Shu-Cherng Fang This paper advances ``Equations for Linearizing Discrete Product’’ — a set of equations that linearize discrete polynomial terms in an effective manner. We exploit the logarithmic feature and propose an enhancement technique to further reduce the number of linear constraints for the current ELDP-based reformulations. We extend the applicability of ELDP to a more general set of ``representable programming problems’’. Computational experiments on both random problems and literature-reported problems support that the proposed method indeed increases the computational effectiveness. 4 - A Farm-level Precision Land Management Model Based On Integer Programming Qi Li, Iowa State University, 0076 Black Engineering, Ames, IA, 50010, United States, qili@iastate.edu, Guiping Hu, Zaki Jubery, Baskar Ganapathysubramanian A farm level precision farmland management model based on mixed integer linear programming has been proposed. Optimal decisions are provided for preseason crop planning and irrigation water allocation. The model captures the effect of size and shape of decision scale as well as special irrigation patterns. We illustrate the model by conducting a case study on a farm in California and show the model is capable of representing the decision making process and flexible to accommodate various scenarios. The results show that a threefold increase of annual net profit could be achieved by carefully choosing irrigation and seed types. 5 - Whistler-Blackcomb Mega Day As An Integer Programming Problem John Shane Lyons, PhD Student, Western University-Ivey Business School, 23 Pine Ridge Drive, London, ON, N5X 3G7, Canada, jlyons.phd@ivey.ca The Whistler-Blackcomb ski area has recently implemented a system which tracks use of 25 lifts across both mountains. A web-based application offers skiers a variety of analytic information pertaining to their skiing activity, and poses a number of ‘badges’ to be earned. One of these, called Mega Day, requires a skier to ride all lifts on both mountains, and consequently ski roughly 10,000 vertical metres. On a particular date of study less than 0.1% of skiers on the mountain achieved this goal. This presentation describes an integer programming model developed to identify feasibility and optimal routing for the Mega Day challenge. Opt, Network III Contributed Session Chair: Oliver Lum, U of Maryland, College Park, 11604 Parkedge Drive, Rockville, MD, 20852, United States, oliver@math.umd.edu 1 - On The Complexity Of Sparse Maximum Flow Problems Andrew Romich, Systems Engineer, Sandia National Laboratories, 4617 Gerrilyn Way, Apt 206, Apt. 206, Livermore, CA, 94550, United States, aromich@sandia.gov, Cole Smith, Guanghui Lan We consider several sparsity-inducing restrictions of the standard maximum flow problem. Each restriction is a unique combination of the number of nodes having an out-degree greater than one and the inclusion of either a restriction on the total amount of flow on all arcs or the total number of arcs used in the network. Assuming an option of initially allowing continuous or integer flow, we analyze a total of twelve problems. TC78 Legends F- Omni

2 - Mathematical Optimization For Managing Selective Catalytic Reduction For A Fleet Of Coal-fired Power Plants Jay Michael Rosenberger, University of Texas-Arlington, Dept of Industrial Manuf Systems Eng, PO Box 19017, Arlington, TX, 76019, United States, jrosenbe@uta.edu, Antonio Alejandro Alanis Selective Catalytic Reduction (SCR) reduces emissions of oxides of nitrogen (NOx) in coal-fired power plants. With a given set of scheduled outages for a fleet of power plants, we develop a heuristic multi-commodity flow problem with schedule elimination constraints, and a knapsack problem with assignment constraints, to create an SCR management plan, which minimizes the total SCR operational cost of the entire fleet of power plants and maintains NOx emissions below a desired target. We present results indicating that our approach provides an equally optimal solution to that of a schedule generation and optimization approach in less computational processing time. 3 - A Parametric Linear Programming Approach To A Cluster Ratio Based Hierarchical Consensus Clustering Algorithm Victoria Ellison, North Carolina State University, 2152 Burlington Labs 2500 Stinson Dr., Raleigh, NC, 27695, United States, vmelliso@ncsu.edu, Yahya Fathi, Amy Langville The minimum cluster ratio cut is a well studied partition-based clustering method used to partition a data set into two or more clusters of balanced size. We propose a natural extension of the minimum cluster ratio cut, which we call the minimum beta-ratio cut, which lends itself to hierarchical clustering methods. We propose a divisive hierarchical consensus clustering algorithm which uses the minimum beta-ratio cut and a heuristic for this algorithm by modifying a BILP formulation of Median Partition problem and applying a parametric programming algorithm to it. 4 - A Hybrid Heuristic For The Windy Rural Postman Problem With a Time-Dependent Zigzag Option Oliver Lum, U of Maryland, College Park, 11604 Parkedge Drive, Rockville, MD, 20852, United States, oliver@math.umd.edu, Rui Zhang, Bruce L Golden In some vehicle routing applications, service is required along both sides of the road. During early morning hours, the vehicle may service both sides of the street during a single traversal by zigzagging between the two. This scenario is modeled by the Windy Rural Postman Problem with Time-Dependent Zigzag Options (WRPPZTW). We present a hybrid heuristic for the WRPPZTW which uses well- known insertion techniques in concert with a new integer programming formulation for the WRPPZ. We compare computational performance relative to an existing exact procedure for a set of small instances and present more extensive results for a set of larger instances. 5 - The Deviation-flow Continuous Location Problem For A Single Refueling Station On A Tree Network Sang Jin Kweon, PhD Candidate, Pennsylvania State University, In this study, we address the continuous version of the single refueling station location problem on a tree network considering that a given portion of drivers has the option to deviate from their preplanned paths to be able to refuel their vehicles. The objective of the problem is to determine the set of optimal station locations that maximizes the total traffic flow covered (in round trips per time unit). To achieve this goal, we derive optimality properties regarding deviation and determine the sets of candidate sites to locate the refueling station. Then, we develop a polynomial algorithm to determine the set of optimal locations. 310 Leonhard Building, Industrial and Manufacturing Engineering, University Park, PA, 16802, United States, svk5333@psu.edu

TC79 Legends G- Omni Opt, Stochastic III Contributed Session

Chair: Md Abdul Quddus, PhD Student, Mississippi State University, Department of Industrial & Systems Engineering, PO Box 9542, Starkville, MS, 39762, United States, mq90@msstate.edu 1 - Multi-depot Fleet Dimensioning For Seasonal Stochastic Demand Marcus V. Poggi, PUC-Rio, Rua M. S. Vicente 225, Rio de Janiero, 22591-900, Brazil, poggi@inf.puc-rio.br, Fabian Penaranda Given a predicted demand behavior represented by a generation function over time and space, distances between clients and the depots, cost to lease one vehicle for the period, unity cost for traveling a distance for leased and for short time hired vehicles and a demand allocation rule, find the (heterogeneous) fleet size at each depot that minimizes the expected operation cost for the period. We propose a decomposition approach based on Stochastic Dual Dynamic Programming that allows dealing with the integrality of the sub-problems. Instances based on CVRPLIB are generated and experimental results are presented.

330