The Austin Chapter of INFORMS announces the second annual Austin INFORMS symposium to be held at the Austin complex of Schlumberger on Friday, February 23, 2001.

Purpose

General Information

Registration

Directions

Preliminary Schedule

Abstracts

Paul A. Jensen
University of Texas at Austin, Graduate Program in Operations Research and Industrial Engineering

"The History of Optimization, Mathematics Driven by Applications"

With a few exceptions, mathematics was not applied to optimization with constraints prior to the Second World War. Then in 1947, in an attempt to solve a complex planning problem for the U.S. Air Force, the first linear programming models were formulated and the simplex algorithm was developed. Since that time the growth in computer power and the increasing complexity of modern society has driven the development of the theory and algorithms for mathematical optimization. Today, powerful modeling and solution techniques are available on our desktop computers. Optimization based decision systems have transformed operations in many industries. Continued developments in computers, mathematical theory and algorithms promise a healthy future.

This talk gives examples of early attempts at modeling and algorithms and surveys the development of this field to the present.
 

KEYNOTE ADDRESS:

Robin Lougee-Heimer
IBM, Yorktown Heights, New York

"Optimization Applications and Innovations at IBM Research"

IBM has a unique distinction within the operations research world. Not only is IBM a prodigious user of operations research in its internal manufacturing operations, the company is also a leading provider of operations research consulting, services, and software products.  Underpinning many of these endeavors is the IBM Optimization Solutions and Library (OSL), a suite of high performance optimization applications and a function library. In this presentation, Dr. Lougee-Heimer will give an overview of the optimization work at IBM, including the advancements in the recently released version of OSL and the new initiative to promote open source software for the operations research community, COIN-OR.

Bio: Robin Lougee-Heimer is a Research Staff Member in the Mathematical Sciences Department at the IBM T. J. Watson Research Center in Yorktown Heights, New York.  As an industrial researcher, Dr. Lougee-Heimer is charged with both conducting basic research and supporting IBM's businesses.  Her main research interest is in developing efficient solutions to large-scale discrete optimization problems arising in industry.  Dr. Lougee-Heimer is a leader in the initiative to promote open source software for the operations research community.  As a  member of the Common Optimization Interface for OR (COIN-OR) core-team, she is a principal infrastructure manager and is leading the development of an open cutting plane library.

Since joining IBM Research in 1993, Dr. Lougee-Heimer has contributed to the core technology, design, and implementation of a variety of decision support systems for IBM manufacturing. Her algorithms and models have been embedded in tools ranging from prototypes that analyze the load balancing of ion implanting machines to production systems that assess the supply-demand capability of IBM's worldwide manufacturing supply chain.

Dr. Lougee-Heimer earned her PhD in Mathematical Sciences from Clemson University in 1993.  She is an active  member of the Institutes of Operations Research and the Management Sciences.  Dr. Lougee-Heimer is a frequent conference organizer and INFORMS Fora officer. Currently, she is serving as an associate editor for IOL, and co-chair of IBM's Operations Research Community.  Dr. Lougee-Heimer is a recipient of two IBM Research Division Technical Group awards, and has published in the areas of mathematical programming, manufacturing production planning, object-oriented application frameworks, and hydroelectric management.
 

David Morton
The University of Texas at Austin, Graduate Program in Operations Research and Industrial Engineering

"A Mandate for Modeling Uncertainty in Optimization Projects"

Optimization algorithms from linear, nonlinear, and integer programming are powerful tools for solving mathematical models of many important real-world planning problems. Deterministic formulations, however, fail to capture the uncertainty inherent in many systems. As a result, solutions to deterministic optimization models can be unsatisfactory as they fail to hedge against future contingencies. Stochastic programming provides an appropriate modeling framework for many such problems, and over the last 10-15 years there has been a dramatic increase in its application. This is primarily due to improvements in computing facilities, methodology, and most recently, software availability. We will discuss several stochastic programming applications, give insight into qualitative differences that can arise in solutions to stochastic and deterministic models, and describe developments that are on the horizon.
 

Leon Lasdon
University of Texas at Austin, Graduate School of Business
Zsolt Ugray
University of Texas at Austin, Graduate School of Business

"Solving Nonconvex NLPs and MINLPs Using OptQuest/GRG"

OptQuest/GRG uses OptQuest, a global optimizer, to provide starting points for GRG, a gradient-based local NLP solver.  GRG seeks a local optimum from a subset of these points, holding discrete variables fixed.  Optquest, developed by Glover, Kelley, and Laguna, uses tabu and scatter search.  Like Genetic algorithms, it maintains and updates a population of good solutions, and uses them to generate new candidate solutions by taking linear combinations of population elements.  GRG starts from each candidate, unless it is too close to a previously found local optimum or has too high a value for an exact penalty function.

Computational results include 140 GAMS problems due to Floudas, et. al.  These include nonconvex quadratic programs, bilinear and biconvex problems, and many problems involving chemical process operation and design.  OptQuest/GRG finds globally optimal solutions to almost all problems solved thus far in a small number of GRG calls.  Important algorithm parameters include total iterations, initial OptQuest calls, and whether or not OptQuest is provided information on linear constraints.  The procedure cannot guarantee a global solution, but promises to find optimal or good solutions to large problems with less effort than competing procedures.
 

Douglas J. Morrice
University of Texas at Austin, Graduate School of Business Schlumberger Austin Product Center

"Optimizing Operations in 3D Land Seismic Surveys"

We formulate and solve a mathematical programming optimization model to find a minimum cost solution for an orthogonal split-spread design in a 3D land seismic survey. The model contains decision variables on source and receiver location spacings, the amount of receiver equipment, and the production rate of the seismic crews. The model includes operational constraints for source and receiver movements. It also includes geophysical constraints for fold coverage, offset, and azimuth. To demonstrate the efficacy of the model, we include an example and solve it using the non-linear optimization solver in Microsoft Excel. The model results demonstrate the classic trade-off between source and receiver points to satisfy the aforementioned geophysical requirements. In addition, we conduct sensitivity analysis on an important production parameter: the maximum number of source points that can be shot per day. We show that while changes to this parameter do not impact the decision variables, such changes do have a significant effect on the total cost of the survey.
 

Dirk Guenther
Sabre, Dallas

"Operations Research Models for Flight Scheduling"

Flight scheduling is a complex task that spans over 12 months. Airlines tackle the problem in several stages using a variety of Operations Research models. Our presentation gives an overview over the models that are currently used, the impact they have had on the scheduling process and their limitations.
 

Maarten Oosten
PROS Revenue Management, Houston

"Implementing Revenue Management Models"

Modeling and solving a revenue management problem using standard modeling and optimization tools is only half of the story for a successful implementation. In this talk, after an introduction to revenue management, we will discuss what is involved in realizing and embedding an optimization solution in a software product.

Javad Ahmadi
Advanced Micro Devices

"Scheduling Optimization at Advanced Micro Devices"

AMD has undertaken a major initiative in deployment of a supply chain planning system to support its needs for a fully integrated planning and visibility system covering demand planning, production planning, scheduling and available to promise functionalities for its extensive domains of production, marketing and distribution.  The Adexa suite of iCollaboration software was chosen for this purpose after significant evaluation of competing offerings with participation of business units in AMD.  Prior to this significant commitment, smaller scale initiatives were undertaken for support of planning activities within AMD.  MPP, a planning tool developed in 1999 for 2000 deployment is an example of such work.  While these tools will be phased out in a timely fashion, they have served invaluable in transition from ad-hoc methods to optimization based methods, user expectation, and understanding.  Many of the developed methods in these areas will be reused as part of the interface, integration and infrastructure development effort required for deployment of the new applications, which is the most cumbersome aspect of such applications delivery.  This presentation describes MPP.
 

Additional Information

http://mohican.me.utexas.edu/~informs/
Austin Chapter of INFORMS
c/o Assistant Professor John Hasenbein
Graduate Program in OR/IE
Dept. of Mechanical Engineering
University of Texas at Austin
Austin, Texas 78712-1063
(512) 471-3079

Organization Contact Information

INFORMS
http://www.informs.org/
90 Elkridge Landing Road, Suite 400
Linthicum, MD 21090-2909
(410) 850-0300
 

Email questions to:
jhas@mail.utexas.edu
Tsmith@trajecta.com