![Show Menu](styles/mobile-menu.png)
![Page Background](./../common/page-substrates/page0357.png)
INFORMS Nashville – 2016
355
TD63
Cumberland 5- Omni
Location Analysis II
Sponsored: Location Analysis
Sponsored Session
Chair: Zvi Drezner, California State University-Fullerton,
800 N. State College Blvd., Fullerton, CA, 92834-3599, United States,
zdrezner@fullerton.edu1 - Optimal Placement Of M Finite-size Rectangular Facilities In An
Existing Layout
Rakesh Nagi, University of Illinois, Urbana-Champaign, 117
Transportation Building, MC-238, 104 South Mathews Avenue,
Urbana, IL, 61801, United States,
nagi@illinois.edu, Ketan Date
We study the problem of placing M new finite size rectangular facilities (NFs) in a
layout with N existing rectangular facilities (EFs). Interactions are present
between different pairs of EFs and NFs, serviced through the Input/output points
located on the facility boundary. The objective is to minimize the total weighted
rectilinear distance between the interacting facilities by optimally placing the NFs.
Main contribution of this paper is an analytical framework that unifies and
generalizes the facility location/layout problems for minisum objective and
rectilinear distance metric.
2 - Solving The Quadratic Assignment Problem Using Graphics
Processing Unit Clusters On The Blue Waters Supercomputer
Ketan Date, University of Illinois at Urbana-Champaign,
Urbana, IL, United States,
date2@illinois.edu, Rakesh Nagi
In this work, we discuss a parallel branch-and-bound algorithm for solving the
Quadratic Assignment Problem (QAP). Our parallel architecture is comprised of
CUDA enabled NVIDIA Graphics Processing Unit (GPU) clusters on the Blue
Waters supercomputer at the University of Illinois at Urbana-Champaign. For
obtaining a lower bound, we adopt the RLT2 formulation of the QAP, and we
propose a novel parallelization of the Dual Ascent algorithm on the GPUs, which
shows excellent parallel speedup for large problems. We show that this GPU-
accelerated approach is extremely effective in solving large QAPs to optimality.
3 - Discrete Budget Allocation In Competitive Facility Location.
Tammy Drezner, Cal State Fullerton,
tdrezner@fullerton.edu,
Zvi Drezner
We apply the gravity-based model for estimating the market shares attracted by
competing facilities. We assume that a budget is available for expanding existing
facilities and building new ones. We assume that the investments for improving
existing facilities or constructing new ones are an integer multiple of a basic value
such as 0.1% of the available budget.
4 - An Iterative Procedure For Solving Non-Convex
Non-Linear Programs
Zvi Drezner, Cal State Fullerton,
zdrezner@fullerton.edu,
Pawel J. Kalczynski
Non-linear programming problems of minimizing a convex objective function
subject to convex constraints are convex, and can be optimally solved by
numerous approaches and canned programs. Non-convex programs such as a
maximization of a convex objective subject to constraints which are outside of
convex regions usually have many local optima and are generally difficult to
solve. We found that such problems can be heuristically solved by a multi-start
approach based on solving a sequence of linear programs. Our iterative approach
is much faster than a direct multi-start approach (one to three orders of
magnitude) and provided better results on four test problems and 116 instances.
TD64
Cumberland 6- Omni
Multiple Criteria Decision Making Applications 2
Sponsored: Multiple Criteria Decision Making
Sponsored Session
Chair: Murat Mustafa Koksalan, Middle East Technical University,
Ankara, Turkey,
koksalan@metu.edu.tr1 - Probabilistic Algorithms For Multiple Criteria Sorting
Sinem Mutlu, Roketsan,
sinemmutlu01@gmail.com,
Murat Mustafa Koksalan, Yasemin Serin
We develop interactive approaches to place alternatives that are defined by
multiple criteria into preference-ordered classes. Our approaches place
alternatives into classes either deterministically or probabilistically with a desired
level of accuracy. We also control the magnitude of misclassification regarding the
number of classes between the true and placed classes of a misplaced alternative.
We demonstrate the approach on a variety of problems.
2 - Properties Of Optimal Stochastic Programming Solutions In
Portfolio Optimization With Different Criteria And
Planning Periods
Ceren Tuncer Sakar, Hacettepe University,
cerents@hacettepe.edu.tr, Murat Mustafa Koksalan
Incorporating multiple criteria, considering different risk measures and using
multiple-period models have been recent important developments in portfolio
optimization. We make a detailed analysis of the properties of optimal stochastic
programming solutions for portfolio optimization problems. We work with
models that have different criteria and planning periods. We compare the
solutions of single and multiple-period models using expected return and
Conditional Value at Risk as criteria, and demonstrate our results with tests
performed with stocks traded on Istanbul Stock Exchange (Borsa Istanbul). We
also consider rolling horizon settings.
3 - An Interactive Approach For Biobjective UAV Route Planning In
Continuous Space
Murat Mustafa Koksalan, Middle East Technical University,
Indus Engineering Department, Ankara, 06531, Turkey,
koksalan@metu.edu.tr, Diclehan Tezcaner Ozturk, Hannan Tureci
We consider the route planning problem for Unmanned Air Vehicles (UAVs)
which we formulate as finding the path that the UAV follows in a continuous
terrain visiting all target points. We consider two criteria: minimization of distance
traveled and radar detection threat. We develop an interactive algorithm that
finds the most preferred point of a route planner (RP). We assume that the RP has
an underlying linear preference function whose parameters are unknown to us.
We ask the RP to compare pairs of tours and his/her preferences guide us to
his/her most preferred nondominated point.
4 - Interactive Algorithms For a Wide Variety Of Preference Functions
Gulsah Karakaya, Middle East Technical University, Ankara,
Turkey,
kgulsah@metu.edu.tr, Murat Mustafa Koksalan,
Selin Damla Ahipasaoglu
In this study, we introduce a broad family of preference functions that can
represent a wide variety of preference functions. We develop interactive
algorithms that guarantee to find the most preferred solution of a decision maker
whose preferences are consistent with such functions. Our algorithms converge to
the most preferred solution of the decision maker by reducing the solution space
based on the preference information obtained from the decision maker and the
properties of the assumed preference functions. We demonstrate the algorithms
on an example problem.
TD65
Mockingbird 1- Omni
Social Media and Health 2.0
Sponsored: Information Systems
Sponsored Session
Chair: Lu Yan, Indiana University, Indiana University, Bloomington, IN,
47405, United States,
yanlucy@indiana.edu1 - How Online Comments And Government Ratings Affect Patients’
Opinion Of Medical Providers
Weiguang Wang, University of Maryland, 3330 B Van Munching
Hall, College Park, MD, 20742-1815, United States,
weiguangwang@rhsmith.umd.edu, Niam Yaraghi,
Guodong (Gordon) Gao, Ritu Agarwal
One critical decision for every patient is to choose a high quality doctor. In recent
years, new online channels have profoundly changed how patients access
physician quality information. Most notably are the government-led efforts such
as PhysicianCompare, and the grass-root movement by voluntary patient reviews
such as those on
Yelp.com. However, little is known how these two channels
affect patient decision making. Using experimental designs, we examine patient’s
choice of primary physicians with quality information from both Yelp and the
government website. Our study provides the first empirical evidence of how
patients weigh different information sources to inform their decision making.
2 - Modeling Dynamics Of Service Mechanism, Feedback
Mechanism, And Sharing Mechanism: An Empirical Analysis
Using Vector Autoregression
Liuan Wang, Harbin Institute of Technology, Harbin, China,
wangliuan1973@163.com, Xitong Guo
With the utilization of social media in healthcare online healthcare communities
has become an integral part of people’s daily lives. In this study, we explore how
the interdependencies among service mechanism, feedback mechanism, and
sharing mechanism affect physicians and patients in the online healthcare
communities. We use vector auto-regression to model the co-movements of
service mechanism, feedback mechanism, and sharing mechanism and provide
evidence of strong Granger-causal interdependencies. In addition, we also
investigate the effect of values in the online healthcare communities. Our results
provide both theoretical and practical implications.
TD65