CPMS, The Practice Section of INFORMS, sponsors workshops on topics of interest to isolated practitioners, i.e. Lone Rangers, of OR and MS. The 11th workshop in this series to be held at INFORMS meetings will address timely issues affecting the grow-ing isolated practitioner community. All are welcome to come and participate in this workshop and discussion Monday, October 27, 6:15-7:15pm in Regency B.
The Business Meeting & Luncheon will be held at lunchtime, Monday in the Cascade B Room. Everyone interested in issues related to women in OR/MS is invited to attend. Bring a bag lunch and join us 12-1:00pm.
The Reception for the INFORMS Forum on Women in OR/MS will be held from 7:00- 9:00pm on Sunday in the Cascade B Room. This is a good opportunity to meet women researchers and practitioners in OR/MS. Everyone is welcome. The reception is sponsored by The Optimization Center at IBM's T.J. Watson Lab.
All participants and their registered guests are invited to attend the INFORMS General Reception on Monday evening, October 27 from 7:30-10:30. The General Reception will be held in the Union Station featuring Tex-Mex food and drink. Don Walser's Pure Texas Band will provide music at the reception from 8:30-10:30pm and we plan to have two-step lessons for members and guests. For admittance, please present the ticket found in your registration packet.
The INFORMS General Membership Meeting will be on Tuesday, October 28, 7:30-8:30pm in the Cascade B Room. Wine and cheese will be offered and the INFORMS Board Members and Officers will provide the entertainment.
The Omega Rho Plenary, "Expanding OR/MS to New Dimensions of Use & Research: From "Ex Ante Planning" to "Ex Post Evaluation & Control" will be presented by William W. Cooper on Tuesday, October 28 in Reunion E-F from 11:30am-12:30pm. Prior to the presentation, Professor Cooper will be inducted as the 26th Honorary Member of Omega Rho.
The program of the Student Activities Committee addresses the social, professional and academic interests of student attendees. An initiative for students, by students, that complements the efforts INFORMS always makes for student attendees.
The program has five parts: a Sunday panel discussion on student chapter concerns; Sunday talks on professional and academic interests, a Sunday evening Student Reception (hosted by INFORMS), a dedicated suite in which students and industry can mingle and social activities to be announced at the meeting.
Special sessions of interest to students will be held on Sunday in Track 26.
For further information, contact any Committee member or visit our website at http://uts.cc.utexas.edu/~bibo: Bruce Colletti, Coordinator, bcolletti@mail.utexas.edu, 512-471-1336; Astrid Kenyon, Professional Program, leuba@posse.me.utexas.edu; Chris Kenyon, Academic Program, cmkenyon@mail.utexas.edu; G. V. Ramanan, Programs Deputy, gvr@mail.utexas.edu; Sobhi M. Mahmassani Publicity and Webmaster, bibo@mail.utexas.edu.
Hosted by INFORMS, this reception is open to all student attendees Sunday, October 26, 7:00-9:00pm in the Cascade A Room. Come network with other students (and others who drop by) in a forum that serves free food and drinks. An optimal way to cap off a long day. For admittance, please present the ticket found in your registration packet.
The Welcoming Session will feature opening remarks by Art Geoffrion, President of INFORMS, and Paul Jensen, General Chair of the Dallas meeting. The following awards and prizes will be given: the Philip McCord Morse Lectureship, the Lanchester Prize, the INFORMS Prize and the INFORMS Expository Writing Award.
Is this your first INFORMS Meeting? Are you new to INFORMS? We would like to welcome you and help you discover the many benefits of INFORMS, and most importantly, how to get the most out of the INFORMS Meeting in Dallas. We will help you to discover things that are valuable to you at the INFORMS Meeting, e.g., technical information, applications, networking opportunities, social opportunities, job opportunities, etc. and how to find them. We will help you find the technical papers, tutorials, workshops, subdivision meetings, receptions, parties, business meetings, free food, exhibits, jobs, wine and cheese, and people. We will answer your questions about our 10 journals, about our individual member benefits such as life, disability, and health insurance, credit card, internship programs, and job placement services. We will introduce you to INFORMS On-Line and your virtual Institute. Do you want to know more about how to get involved in the Institute? Do you know about the Chapters in your area, or the many sections and colleges that focus on a specific methodological or application interest? We will provide the answers to these questions and more. This workshop will also give you a chance to meet some INFORMS leaders, to learn more about the Institute, and to get answers to all of your questions about the Institute.
Whether you are a new member or an experienced INFORMer, you will learn something new about INFORMS Meetings and the Institute. (It is also a good way to get a discount coupon for many of the Institute's products.) The workshop will be held on Sunday, October 26, 10:30am-12noon in the Duncan A Room. We look forward to seeing you.
In December 1995, Hewlett-Packard acquired Convex Computer Corporation, located in Dallas, Texas, to extend its technical computing solutions from the desktop to scalable parallel-processing systems. HP is now the only vendor offering scalable, binary-compatible, technical computing solutions with shared memory and message-passing capabilities, from desk-tops to supercomputers.
While touring the Convex Division of HP, you will see a modern manufacturing facility utilizing state-of-the-art manufacturing and material management strategies. Software systems are utilized for product tracking, quality data entry, on-line documentation generation and documentation management. The factory has a modern circuit board assembly process, electro-mechanical assembly and a system configuration and test operation. Minimum material storage is present due to the utilization of demand flow technology in the material and supplier strategies.
The tour will visit 3 areas: the first part of the tour will be the final assembly line for the Army's OH-58D Kiowa Warrior armed reconnaissance helicopter and the Marines' AH-1W SuperCobra attack helicopter. Here you will see how these helicopters are assembled and the high-tech equipment used. Bell's final assembly line is claimed to be world class. From here you will tour the composite area where parts for the V-22 Osprey tiltrotor aircraft are made, including some radically new computer-driven manufacturing devices. Finally, you will visit Bell's Hall of Heritage where Bell's history is displayed in photos, videos, models and artifacts. The Hall highlights Bell's initial move in vertical flight in the early 1940s.
The middle/high school and community college math and science teachers program will be composed of introductory OR and computer application workshops. An introduction to OR/MS will be presented, followed by probability modeling, computer simulations, queueing or waiting line models and linear program-ming. Attendees are welcome to attend the general conference and each will receive copies of all program materials: videos, a subscription to OR/MS Today and copies of all software packages with instructions and teacher/student activity guide. The program will be held on Monday, October 27, 8:00am - 5:00pm in the Brisbane B Room.
Analyses conducted in various parts of an enterprise often produce conflicting or disconnected results. This becomes especially true as an organization grows in size. Until a more holistic sense of decision making can be developed, decision makers will continue to make decisions by the seat of their pants. The speaker will share his experiences bringing operations, technology and economic analyses together to forge a cohesive picture for sound business decisions at the FAA.
The airline industry is currently actively involved in developing and implementing mathematics programming models in planning. Three such problems are considered: crew pairing, fleet assignment, and bid-prices. These problems are currently being solved regularly by major airlines using mixed-integer (for the first two) and non-linear programming (for the third). We discuss size of models, hardware and software being used, and some of the methodology used to overcome computational difficulties. The crew pairing and bid price problem involve solving LPs with many, many columns. Some general approaches to such problems will be given. Large-scale decomposition remains an active area in trying to develop integrated models. Both modeling and computational issues will be discussed.
The field of OR has made tremendous impact in managing and controlling today's airline operations. Accelerated by modern computing technologies, based on improved understanding of the business logic applied in airline industry, enabled by advancement of optimization models and solution techniques, driven by enormous demand from airlinesí management for gaining competitive edge in the market, the real-time, mission-critical DSSs for managing and controlling airline operations start to become a reality. This tutorial discusses decision support system applications in airline real time operations.
The goal is to present in a rather concise format what needs to be known in order to design and implement an efficient Lagrangean relaxation in integer programming. The presentation will start with a basic introduction to Lagrangean relaxation for linear integer problems and its geometric interpretation. General ideas for splitting a problem before applying Lagrangean relaxa-tion will be presented. After a brief review of the important characteristics of the Lagrangean function, we will turn to primal and dual methods for solving relaxation duals: the classic subgradient optimization and constraint generation methods and a more recent, hybrid, 2-phase approach.
Projection and inverse projection are twovery simple, yet unifying, concepts in the theory of linear inequalities. Using projection, we take a system of linear inequalities and replace some variables with additional constraints. Inverse projection, the dual of this process, involves replacing inequalities with variables. We extend the inverse projection concept to the case where some or all of the variables are required to be integer. This process yields extended polyhedral representations in auxiliary variables. We describe five methods for generating reformulations with auxiliary variables.
The task of building efficient implementations of combinatorial algorithms presents a number of interesting challenges, distinct from those that arise in the study of the worst case complexity of the algorithms. We present a survey of results in this area, concentrating on network optimization problems, including minimum-weight matchings, maximum flow, minimum cut, minimum-cost flow, minimum-weight spanning tree and the traveling salesman problem. We will consider exact algorithms, heuristic algorithms and cutting-plane methods. In each case, the target will be the development of implementations that scale up to problem instances having one million or more nodes.
We review recent theoretical and empirical progress in the development of approximation algorithms for hard scheduling problems. In particular, we discuss the use of LP-relaxations in deriving performance guarantees, a technique for identifying limits of approximability and a variety of neighborhood search approaches for finding reasonable schedules reasonably fast.
This tutorial describes the theory and the design of algorithms related to the use of the reformulation/linearization/convexification technique. For discrete linear and polynomial programming problems, this technique generates a hierarchy of progressively tighter, higher dimensional, linear programming representations, leading to an explicit characterization of the convex hull of feasible solutions. Several applications that arise in location-allocation, distribution and engineering design contexts have been studied using this approach and the results indicate that this technique can often recover near optimal solutions via a single or a few linear or convex programming relaxations.
Downsized and flatter organizations, exploding volumes of data and rapidly changing market-conditions have combined with powerful new software and personal computers on networks to produce problems and opportunities. The problem is that the traditional marketer's job has been vastly expanded, from one that has historically relied predominantly on intuitive and qualitative skill to one that now has great quantitative demands as well. The opportunity is that the new hardware and software, combined with proper training, can turn conceptual marketers into marketing engineers who apply the OR/MS approach to marketing problems.
Until recently, our required "Introduction to OM/OR" course was subtitled "Course from Hell taught by Anti-Christ" by the students. In contrast, many of our current students think this is the best course they took so far in the Faculty of Business. We will describe the components of this metamorphosis: extensive use of spreadsheets, demos, labs, electronic communication, cases, games, music, skits, and stupid professor tricks. We believe that many of these catalysts are portable to other institutions.
This tutorial will attempt to help the OR/MS instructor with the teaching of OR/MS cases. Topics will include: What is a "case"? Why use cases in an OR/MS course? Where do I find OR/MS cases? Choosing appropriate cases for your course. Preparing for a case class. Strategies and tactics in the classroom.
Classes in introductory probability are usually taught topically oriented. Students are taught the mechanics of probability to manipulate well formulated problems. This tutorial provides a teaching framework to bridge the gap separating mechanics and problem formulation/communi-cation. We will share teaching materials (including software) developed over many years with many iterations of modifications (both radical and fine-tuning).
The power of DEA for analyzing operational processes has been demonstrated by numerous applications including a recent finalist for the Edelman Award. However, in spite of this exposure, the methodology remains unfamiliar to significant portions of the OR/MS community. This tutorial will describe what DEA is conceptually and show why/ how to perform an analysis.
This tutorial presents the current state-of- the-art in using visualization to support all aspects of the modeling process, from model and algorithm development through to final results presentation. Topics covered include animation, virtual reality, sound, hypermedia, among others. The tutorial includes extensive, live examples.
The emergence of electronic commerce as an active commercial area has opened up many new research challenges. This tutorial will outline some of the topics and suggest approaches to their solution. Successful solution will draw upon the fields of computer science for technology and economics to describe the business value. Some examples are in real-time network management, competitive analysis of electronic-based companies and real-time demand estimation for digital products. Tools and techniques, including electronic cash, authentication and security, will be outlined.
Designing and training neural networks using linear programming raises an important set of questions about how the brain actually learns. It also challenges some of the basic notions of learning used to develop algorithms in the neural network field. This tutorial will survey linear programming-based methods to generate neural networks for supervised and unsupervised learning. Robustness and reliability of linear programming-based methods over existing neural network methods will be discussed and an overview of the various applications will be provided.
We overview the salient developments for modeling and assessing the reliability of computer software. The models that seem to be promising are based on point processes, and inferential issues invariably involve the Bayesian paradigm with Markov chain Monte Carlo methods. We discuss topics such as optimal testing and classification.
A classic and recently popular method for solving many hard combinatorial problems is based on formulating these problems as set-partitioning models. We explain the excellent empirical performance of the set-partition-ing technique for a number of problems including vehicle routing, bin-packing and parallel machine scheduling problems. We then apply the set-parti-tioning method to tactical problems in supply-chain management involving inventory and transportation. In particular, we consider integrating inventory policies and transportation strategies so as to minimize system-wide costs by taking advantage of the shipping cost structure.
This tutorial reviews the various modeling and solution methodologies for the vehicle routing problem with time window constraints. On the modeling side, we begin with the classic formulation and continue with a multi-commodity flow formulation for multiple depots and multiple vehicle type problems. In terms of solution approaches, we present the different branch and bound and mathematical decomposition methods proposed to solve this problem optimally. Advances in heuristic solutions are also discussed. We then illustrate how to handle issues such as multiple capacity restrictions or time windows, and different types of synchronization constraints.
The electric power industries of the world are now in an unprecedented period of change. One by one, they are being transformed from cost-conscious, regulated utilities to profit-conscious, competitive corporations. Not only does this constitute an about face in emphasis from the production side to the consumer side of the business, but it presents a wonderful opportunity for the use of OR.
New business problems are being created with each legislative act to deregulate and there is a strong need for OR analysts to fill in the gaps. Further, the particular training and experience of our discipline makes us well equipped to absorb, understand and propagate the wealth of new techniques that will be needed for utility companies to succeed in this new environment.
This workshop reviews the latest international developments and explores new areas for research and development that these changes afford us. We will introduce a new set of simple yet demonstrative models that cover the essential aspects of the new business processes arising in our public utilities.
The fact that one can do OR in spreadsheets is well-known and is receiving extensive promotion. However, learning how to teach using spreadsheets (and how to do MS well in spreadsheets) is a non-trivial challenge. We discuss both the gains and the losses of teaching in spreadsheets, the pitfalls to be anticipated and managed when making the transition and the differences between a spreadsheet vs. an algebraic representation of OR models. We share lessons learned regarding the needs of students and challenges and opportunities for instructors when teaching with spreadsheets. Numerous examples will be presented. Although focused towards teachers in business schools, this workshop will be of value to anyone interested in learning about performing OR/MS in spreadsheets.
The verification, validation & accreditation (VV&A) of models and simulations is a topic of high interest in the community. Given the intended use of a model or simulation, it is sometimes critical that it is an appropriate representation of reality. For example, DoD directives emphasize the need to build reliable and credible models and simulations, especially when their results are used to make acquisition decisions. While the motivation for VV&A is clear, it is difficult to prescribe a "cook book" set of techniques and procedures - and for good reason: sponsors, developers and users of models and simulations face different problems involving intended use, risk, model detail/complexity/ size and the availability of real-world data.
This 3-part workshop will focus on the "how to" of VV&A and is aimed at model managers, developers, evaluators and users faced with the challenges of VV&A. Prior experience with statistics is not needed to understand the session on experimental design. Participants should leave with good ideas on how to approach their VV&A journeys and tools to help along the way.
There is much that OR/MS can do to improve the quality of science within maintenance and thereby, impact upon company objectives. This presentation aims to raise awareness of what is now possible by way of modeling support for maintenance and to address the problems of and the potential for the growth of this awareness among both the maintenance and the OR/MS communities.
The national railroad of France, Societé Nationale des Chemins de Fer Français (SNCF) and Sabre Decision Technologies teamed up to address the problem of train capacity allocation in SCNF's network of high-speed electric-powered Trains Grand Vitesse, which carry more than 50 million passengers per year among 140 cities in both France and Europe. The implementation of their resulting DSSs is credited with an overall profit increase of 3-5% reductions in manpower and schedule development time and improvements in planning and scenario evaluation.
As INFORMS enters the 21st century, more and more the emphasis will need to be on the integration of quantitative methods with advances in digital communication, processing and data mining through integrated managerial and operations decision-making. Paperless processing of data and knowledge in a global context will provide INFORMS members the opportunity to extend their attention into real time controls breakthroughs for both planning and control.
More attention to empirical/inferential research, based on ex-post data, is needed to expand OR/MS from its present (almost exclusive) emphasis on ex-ante planning and forecasting of future behavior. Movement in these directions will open new opportunities for uses of OR/MS in "control" as well as "planning" aspects of management. Research pointed to new and better methods for treating data and making inferences will be needed if such opportunities are to be fully exploited. DEA is an example which has provided a start, but more is needed. Some of the likely problems to be encountered and new opportunities for research and use that can thereby be opened will be illustrated by past and present developments in DEA.