2016 INFORMS Annual Meeting Program

WC80

INFORMS Nashville – 2016

3 - A New Mathematical Formulation For Choice-based Optimization Problems Shadi Sharif Azadeh, Ecole Poloytechnique Federale de Lausanne, 38 Route de la Condemine, Lausanne, 1030, Switzerland, shadi.sharifazadeh@epfl.ch, Meritxell Pacheco, Michel Bierlaire The mathematical modeling of choice behavior has been an active field of research. Their complexity leads to mathematical formulations that are highly non convex in the explanatory variables especially when they are integrated inside an MILP to take into account both supply and demand constraints. We propose a new mathematical modeling that can simplify nonlinear and nonconvex choice-based optimization models with the help of simulation inside an MILP framework. We have tested the model on a real case study and computational results testify the goodness of this modeling approach. 4 - Microsoft Excel Evolutionary Solver And Resource Constrained Project Scheduling Norbert Trautmann, Professor, University of Bern, FM Quantitative Methoden, Schuetzenmattstrasse 14, Bern, 3012, Switzerland, norbert.trautmann@pqm.unibe.ch, Mario Gnägi We discuss how to apply the evolutionary solver contained in Microsoft Excel’s Solver Add-in to the resource-constrained project scheduling problem (RCPSP). Combining a novel spreadsheet-based implementation of an appropriate schedule-generation scheme with the evolutionary solver provides surprisingly good schedules. 5 - An Exact Algorithm For The Demand Constrained 0-1 Knapsack Problem The demand constrained KP is a variant of the binary KP in which a weighted summation of the variables must exceed a given threshold in addition to the standard KP constraint. The first exact algorithm for solving this variant is presented which utilizes a reduction routine prior to a breadth-first expanding core approach for determining the remaining variables. A polynomial time solution to the continuous relaxation is employed such that high quality, integer Lagrangian and surrogate relaxations are solved to obtain tight upper bounds. Performance is tested using computational experiments with results demonstrating that the algorithm can outperform commercial software. Chair: Philip F. Musa, Associate Professor and Programs Director, The University of Alabama at Birmingham (UAB), P.O. Box 55544, Birmingham, AL, 35255, United States, musa@uab.edu 1 - Micro-decision Patterns In Prescription Data: An Investigation Of Local And Non-communicable Diseases Among Vertically Differentiated Social W. Art Chaovalitwongse, University of Arkansas, Fayetteville, AR, Contact: artchao@uark.edu, Praowpan Tansitpong, Apirak Hoonlor This study explores electronic healthcare database in defining decision patterns in prescribed medicines among government-subsidized benefit schemes in Thailand. The analysis focuses on three major non-communicable diseases (diabetes, can- cer, and cardiovascular) and three local diseases located in inpatient and outpa- tient database. The study separates uniform prescription patterns in all three schemes from non-uniform patterns and predicts brand and amount of dosage to be prescribed to other diseases. The findings also suggest prescription priority in medicine inventory control. 2 - Minimizing Radiology Error By Improving Staff Scheduling Mahdi Nasereddin, Penn State- Berks, Tulpehocken Road, P.O. Box 7009, Reading, PA, 19610-6009, United States, mxn16@psu.edu, Michael Bartolacci, Michael Bruno Errors are sometimes made when reading radiology charts. A 2001 study reported that depending on the area, the radiology error rate is between 2 - 20%. A team of researchers at the Penn State University is currently studying how to minimize radiology error rate. In this study, the relationship between under-staffing and radiology errors is being investigated. 3 - Evaluating Policy Options For Improving Access To Dental Care For Children In Georgia Benjamin Johnson, Georgia Institute of Technology, 3245 Wellbrook Drive, Loganville, GA, 30052, United States, benjohnson@gatech.edu, Nicoleta Serban, Paul Griffin, Susan Griffin Policies regulating dental providers differ by state. Policies in Georgia are compared to similar policies in other states to estimate the impact of each policy Christopher John Wishon, PhD Candidate, Arizona State University, Tempe, AZ, United States, cwishon@asu.edu, J. Rene Villalobos WC80 Broadway E- Omni Health Care, Public III Contributed Session

on access to dental care. Policy changes are then evaluated to show the impact a new policy could have on improving access to children in Georgia. 4 - Obesity In Africa: The Ultimate Black Belt Region Bursting At The Seams Philip F. Musa, Associate Professor and Programs Director, The University of Alabama at Birmingham (UAB), P. O. Box 55544, Birmingham, AL, 35255, United States, musa@uab.edu What are the root causes of obesity as a Public Health epidemic across Africa? We present comparisons of demographics of those most predisposed to this precursor to chronic comorbidities in developed countries such as the United States and the least developed regions such as Sub-Saharan Africa. It was only fairly recently that mortality rates due to chronic diseases surpassed those due to infectious diseases in industrialized countries. While that has not yet occurred in Africa, there is evidence that the burden due to chronic diseases associated with obesity may soon eclipse those affiliated with infectious diseases. Public Health interventions are suggested in this paper. WC81 Broadway F- Omni Opt, Integer Programming I Contributed Session Chair: Vishnu Vijayaraghavan, Texas A&M University, 400 Nagle Street, College Station, TX, 77840, United States, vishnunitr@tamu.edu 1 - A Branch-and-cut Algorithm Using Two-period Relaxations For Big-bucket Lot-sizing Problems Kerem Akartunali, Senior Lecturer, Strathclyde Business School, Dept. of Management Science, University of Strathclyde, Glasgow, G4 0GE, United Kingdom, kerem.akartunali@strath.ac.uk, Mahdi Doostmohammadi, Ioannis Fragkos We study the polyhedral structure of the two-period subproblem proposed by Akartunali et al. (2016). Based on two relaxations of the subproblem, we propose new families of valid inequalities and present facet-defining conditions. These inequalities are lifted to the original space of the two-period subproblem, and they also inspire the derivation of a new family of inequalities defined in the original space. We exploit the structural similarities of the different families in order to design an efficient separation algorithm, and embed it in a modern branch-and-cut solver. 2 - Prescriptive Analytics To Improve E-warehouse Operations Fatma Gzara, Associate Professor, University of Waterloo, 200 University Avenue W, Waterloo, ON, N2V 2N1, Canada, fgzara@uwaterloo.ca We use data for an e-commerce warehouse characterized with high order volumes, significant seasonality, and a large number of SKUs. Based on extensive descriptive data analysis, the packing operation is identified as a major cause for long order completion times. We develop an optimization model and solution methods based on decomposition and heuristics to optimize order packing. We validate our results using industry data. 3 - Finding Optimal Solutions For Emergency Evacuation By A Dynamic Programming Approach Based On State-space-time Network Representation Lei Bu, Institute for Multimodal Transportation, Jackson, MS, United States, leibu04168@gmail.com, Feng Wang, Xuesong Zhou Based on a representation of state-space-time network, a formulation is proposed to optimize dynamic vehicle routes strategy in an emergency evacuation. The proposed integer linear programming formulation could effectively build the modeling representation of time status, evacuation demand, node capacity and traffic volume change constraints through a multi-dimensional network with an objective function to minimize total travel cost of visiting all nodes. Bellman- Held-Karp algorithm working as a dynamic programming algorithm is utilized to solve the problem. Two scale levels of networks in Mississippi State are tested using the model and algorithm proposed to verify the effectiveness. 4 - A Modified Cutting Plane Algorithm For Inverse Mixed Integer Linear Programming Problems Vishnu Vijayaraghavan, Texas A&M University, 400 Nagle Street, College Station, TX, 77840, United States, vishnunitr@tamu.edu, Kiavash Kianfar, Andrew J Schaefer Given a feasible solution to an optimization problem, the purpose of an inverse optimization problem is to minimally perturb the cost vector to make this feasible solution an optimal one. Inverse optimization for mixed integer programming is particularly challenging and previously a cutting plane algorithm has been proposed. In this paper we present a modification of this algorithm which significantly reduces the number of iterations, and hence run-time, until termination.

452