GUEST LECTURER
Bradley Jones

AFFILIATION
Senior Manager of Statistical R&D, SAS Institute Inc.

ABSTRACT
In industrial experimentation there is growing interest in studies that span more than one processing step. Convenience often dictates restrictions in randomization in passing from one processing step to another. When the study encompasses three processing steps, this leads to split-split-plot designs.

In this talk I will show how to compute D-optimal split-split-plot designs and provide illustrative examples using a pre-release version of JMP software. I conclude by considering D-optimal alternatives to a previously run split-split-plot design for cheese production.

Symposium speakers include:
Cindy Barnhart
Bob Bixby
Bill Cook
Gerard Cornuejols
Marshall Fisher
Rob Garfinkel
Ralph Gomory
Martin Groetschel
John Jarvis
Tom Magnanti
Bill Pulleyblank
Don Ratliff
David Ryan
Mike Trick
Lawrence Wolsey

Thursday, July 26, 2007
1:00 PM - 5:00 PM Scientific and Special Presentations
5:00 PM - 7:00 PM Reception
7:00 PM - 10:00 PM Banquet

Friday, July 27, 2007
8:30 AM - 1:00 PM Scientific and Special Presenations

Support for this symposium is gratefully acknowledged from:
-Center for Electronic Markets and Enterprises (CEME), Robert H. Smith School of Business, University of Maryland
-H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology
-Supply Chain and Logistics Institute, Georgia Institute of Technology

This workshop provides a forum to discuss timely and important issues in energy, environmental science, and logistics researches. Topics include energy policy, energy risk management, environment conscious engineering, environmental cost of economic growth, green GDP accounting, health cares informatics as well as general issues in the logistics and supply chains areas.

We are delighted to invite researchers from the international academic communities to join us in enlightening discussions. The workshop consists of formal presentations which are designed to serve as catalysts for discussion.

Sponsors:
H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology

Management Science and Engineering,
Guanghua School of Management, Peking University

INFORMS Energy, Natural Resource, and Environment (ENRE) Section Contact person:
Related site: http://www2.isye.gatech.edu/irms2007/

GUEST LECTURER
Lori Hougtalen

ABSTRACT
Increasingly, cargo carriers are entering into collaboration in order to improve profitability. A key issue to be resolved when carriers work together in such a manner is how to distribute profit among participating members. We seek to establish a mechanism for allocating profits among cargo carriers who choose to collaborate by sharing capacity on all or a portion of their network. The mechanism should achieve a collaborative system in which (1) system profit is maximized, (2) in the absence of a centralized decision-maker, individual carriers are motivated to make decisions that achieve maximum system profit, and (3) allocations dictated by the mechanism are implementable. Critical components in the design of a mechanism to allocate system profits are the underlying mathematical models. How do we mathematically represent the interests of the system and of individual carriers within the system? We analyze two different modeling approaches and show how differences in modeling lead to significant differences in the properties of the resulting allocations. We find that one model is more flexible in terms of the range of allocations obtained under the model, and we explore how this flexibility may be used to strive for different measures of fairness. We also discuss complications that arise from this flexibility, namely the existence of a secondary market for capacity, which benefits some carriers in the collaboration, at the expense of others. Pizza and refreshments sponsored by The Supply Chain & Logistics Institute.

GUEST LECTURER
Baris Ata

AFFILIATION
Assistant Professor of Operations Management, Managerial Economics and Decision Sciences Department, Kellogg School of Management, Northwestern University

ABSTRACT
(Joint work with Mustafa Akan)

We consider a continuous-time model of network revenue management. Under mild assumptions, we first show that there exists an optimal generalized bid-price control where the bid-price process forms a martingale and is used in conjunction with a permissible capacity process. We also discuss its connection to the bid-price controls in the Talluri &; van Ryzin (1998) sense, and provide sufficient conditions for the (near) optimality of the latter. Second, we construct an epsilon-optimal bid-price control, where the associated bid-price process forms a martingale. Moreover, the epsilon-optimal bid-price control can be viewed as a perturbation of a bid-price control in the Talluri &; Van Ryzin (1998) sense. Third, we show how an epsilon-optimal bid-price control and the corresponding solution to the continuous-time network revenue management problem can be characterized as a solution to a Forward-Backward Stochastic Differential Equation. Finally, we discuss the important special case of continuous information, where one can employ the machinery of Ito calculus.

Bio:
Baris Ata is an Assistant Professor of Operations Management at Kellogg School of Management, Northwestern University, where he has been since 2003. He received BS in Industrial Engineering (1997) from Bilkent University, Ankara, Turkey, and MS in EES &; OR (1999), Business Research (2000), Mathematics (2001) and Statistics (2002), and Ph.D. in Business Administration (2003) from Stanford University. His research focuses on two areas: dynamic control of manufacturing, service and communication systems; and revenue management.

GUEST LECTURER
Prof. Anatoli Iouditski

AFFILIATION
University Joseph Fourier, Grenoble, France

ABSTRACT
The subject of this talk is a complexity analysis of a family of large-scale stochastic approximation algorithms. The methods belongs to the family of primal-dual descent algorithms, introduced by Yu. Nesterov. We propose an adaptive choice of the gain sequences of the algorithm which make it possible to attain the optimal rates of convergence on wide classes of problems. We show, for instance, that if it is known a priori that the objective function is Lipschitz, the proposed algorithm attains minimax rate of convergence. Further, if the objective belongs to a "better class" of smooth functions with Lipschitz-continuous gradient, the proposed algorithm also attains the minimax rate.

GUEST LECTURER
Professor Chang W. Kang

AFFILIATION
Hanyang University, Ansan, Korea

ABSTRACT
Monitoring variability is a vital part of modern statistical process control. The conventional Shewhart R and S charts address the setting where the in-control process readings have a constant variance. In some settings, however, it is the coefficient of variation, rather than the variance, that should be constant. For example this setting is common in clinical chemistry, and then conventional R and S charts cannot be used. This paper develops a chart, equivalent to the S chart, for monitoring the coefficient of variation using rational groups of observations. (This is joint work with Man S. Lee and Young J. Seong of Hanyan University and Douglas M. Hawkins of the University of Minnesota)

GUEST LECTURER
Ton Dieker

AFFILIATION
University College Cork, Ireland

ABSTRACT
Queues in series, combinatorial correspondences, and determinants
(partly based on joint work with Jon Warren, University of Warwick)

There are deep connections between queues in series, noncolliding stochastic processes, and eigenvalues of random matrices. At the heart of these connections lies a combinatorial correspondence ascribed to Robinson, Schensted, Knuth, and Burge. This talk presents some of the remarkable consequences of this correspondence. In terms of queues in series, we give a complete probabilistic description of certain associated stochastic processes. Determinants of matrices play a pivotal role in this description.

GUEST LECTURER
Joseph T. Wu, Ph.D.

AFFILIATION
ISyE - Georgia Tech

ABSTRACT
The outbreak of highly pathogenic H5N1 influenza in domestic poultry and wild birds has caused global concern over the possible evolution of a novel human strain. If such a strain emerges, and is not controlled at source, a pandemic is likely to result. Health policy in most countries will then be focused on reducing morbidity and mortality. We estimate the expected reduction in primary attack rates for different household-based interventions using a mathematical model of influenza transmission within and between households. We show that, for lower transmissibility strains, the combination of household-based quarantine, isolation of cases outside the household, and targeted prophylactic use of anti-virals will be highly effective and likely feasible across a range of plausible transmission scenarios. For example, for a basic reproductive number (the average number of people infected by a typically infectious individual in an otherwise susceptible population) of 1.8, assuming only 50% compliance, this combination could reduce the infection (symptomatic) attack rate from 74% (49%) to 40% (27%), requiring peak quarantine and isolation levels of 6.2%and 0.8% of the population, respectively, and an overall anti-viral stockpile of 3.9 doses per member of the population. Although contact tracing may be additionally effective, the resources required make it impractical in most scenarios. National influenza pandemic preparedness plans currently focus on reducing the impact associated with a constant attack rate, rather than on reducing transmission. Our findings suggest that the additional benefits and resource requirements of household-based interventions in reducing average levels of transmission should also be considered, even when expected levels of compliance are only moderate.

GUEST LECTURER
Professor Ming-Yen Cheng

AFFILIATION
Department of Mathematics, National Taiwan University

ABSTRACT
Data truncation is a problem in scientific investigations. So far, statistical models and inferences are mostly based on the assumption that the survival and truncation times are independent, which can be unrealistic in applications. In a nonparametric setting, we discuss identifiability of the conditional and unconditional survival and hazard functions when the survival times are subject to dependent truncation, namely, the survival time is dependent on the truncation time. A motivating example is discussed. Nonparametric kernel estimators of these unknowns are proposed. Usefulness of the nonparametric estimators are demonstrated through their theoretical
properties, an application to real data and a simulation study. This is a joint work with Peter Hall and You-Jun Yang.

GUEST LECTURER
Joel A. Tropp

AFFILIATION
The University of Michigan

ABSTRACT
A fundamental problem in applied mathematics, statistics, and electrical engineering is to solve underdetermined systems of linear equations. Basic linear algebra seems to forbid this possibility. But a recent strand of research has established that certain underdetermined systems can be solved robustly with efficient algorithms, provided that the solution is sparse (i.e., has many zero components). This talk provides an overview of these sparse representation problems, and it describes the basic algorithmic approaches. Then it details situations where the algorithms are guaranteed to succeed. In particular, the talk introduces some new work on the case where the matrix is deterministic and the sparsity pattern is random. It also covers some results for the case where the matrix is random, which is the situation in Compressed Sensing applications.

GUEST LECTURER
Professor Roger J-B Wets

AFFILIATION
University of California, Davis

ABSTRACT
We consider contingent claims pricing model with continuous distributions, discrete and continuous time. By formulating such problems as stochastic optimization problems and resorting to stochastic optimization and variational convergence techniques, one can obtain constructive procedures in situations that go much beyond the classical Black-Scholes framework. In the process, one derives some important properties such as no-arbitrage and nearly no-arbitrage conditions, equivalent martingale measures, equilibrium equation, etc. Implementation of this strategy requires an, as good as possible, estimate of the price process. A novel estimation approach, that relies on Levy's characterization of a Wiener process, is suggested and implemented experimentally.

GUEST LECTURER
Dr. Philippe Rigollet

AFFILIATION
School of Mathematics, Georgia Tech

ABSTRACT
Several statistical problems where the goal is to minimize an unknown convex risk function, can be formulated in the general framework of stochastic convex optimization. For example, density estimation, regression and convex classification can be treated using the machinery of stochastic optimization. We describe a family of general algorithms called "mirror averaging algorithms" that yields and estimator (or a classifier) which attains optimal rates of convergence in several interesting cases. These optimal rates are illustrated on several examples and compared to standard estimators or classifiers.

GUEST LECTURER
Dr. Nagi Gebraeel

AFFILIATION
University of Iowa

ABSTRACT
Unexpected failures of engineering systems are major contributors to human fatalities and result in astounding maintenance costs. It is estimated that US industry spends $200 billion each year on reliability and maintenance. The high degree of uncertainty associated with degradation processes (even for similar components operating under similar environmental conditions), and our limited understanding of the physics-of-failure are major obstacles in accurately assessing reliability measures and predicting unexpected failures.

This seminar discusses the development of a sensor-based stochastic degradation modeling framework that combines conventional reliability formalisms with condition/health monitoring methods to improve failure predictability. Many physical degradation processes that occur prior to failure can only be observed indirectly using condition/health monitoring techniques that capture degradation-based sensory information. This research rests on the idea that the functional forms of degradation-based sensory signals are driven by the underlying physical degradation phenomena that occur prior to failure. The approach used in this research is to model the evolution of degradation signals using continuous-time continuous-state stochastic models. These stochastic models combine two sources of information; (1) population-specific reliability and degradation characteristics, which are used to estimate preliminary residual life distributions; and (2) in-situ component-specific degradation signals, which are used to update the residual life distributions, in real-time, based on the latest degradation states of the systems being monitored. These dynamically evolving degradation-based residual life distributions will then be utilized to improve reliability assessment and enable the development of sensor-driven replacement and spare parts inventory decision models.

GUEST LECTURER
Yuri Nesterov

AFFILIATION
CORE, Belgium

ABSTRACT
In this talk we discuss a recent progress in the general second-order minimization schemes related to the cubic regularization of the Newton's method. For convex case, we present an accelerated multistep version of the method. We consider the extensions of the new schemes onto constrained problems. Preliminary computational results are also discussed.

GUEST LECTURER
Fei Liu: Faculty candidate

AFFILIATION
Duke University

ABSTRACT
Functional data analysis (FDA) -- inference on curves or functions -- has wide application in statistics. An example of considerable recent interest arises when considering computer models of processes; the output of such models is a function over the space of inputs of the computer model. The output is functional data in many contexts, such as when the output is a function of time, a surface, etc. A nonparametric Bayesian statistics approach, utilizing separable Gaussian Stochastic Process as the prior distribution for functions, is a natural choice for smooth functions in a manageable (time) dimension. However, direct use of separable Gaussian stochastic processes is inadequate for irregular functions, and can be computationally infeasible in high dimensional cases. In this talk, we will develop and extend several Bayesian FDA approaches for high dimensional irregular functions in the context of computer model validation, tailored to interdisciplinary problems in engineering and the environment.

GUEST LECTURER
Lily Wang: Faculty candidate

AFFILIATION
Michigan State University

ABSTRACT
Application of non- and semi parametric regression techniques to high dimensional time series data have been hampered due to the lack of effective tools to address the ''curse of dimensionality''. Under rather weak conditions, we propose a spline-backfitted kernel estimator of the component functions for the nonlinear additive time series data that is both computationally expedient so it is usable for analyzing very high dimensional time series, and theoretically reliable so inference can be made on the component functions with confidence. Simulation experiments have provided strong evidence that corroborates with the asymptotic theory. Finally, the estimation procedure has been illustrated by a US unemployment rate example.

Keywords and phrases: Bandwidths, B-spline, knots, Nadaraya-Watson estimator, nonparametric regression

GUEST LECTURER
Eduardo Izquierdo: Faculty candidate

AFFILIATION
University of Michigan

ABSTRACT
Variation reduction is an important but challenging task in Multistage Manufacturing Processes (MMPs). As an example, dimensional variation of automobile bodies may lead to assembly, aestethic, warranty and safety problems. Therefore, variation should be minimized whenever possible. In this talk I will introduce my research on active compensation of deviation and the design of MMPs focused on variation reduction. Through my talk I will demostrate how those research areas can be integrated. The first part of the presentation will focused on the development of a methodology for active compensation of deviations in MMPs. In particular, I will present the derivation of a feedforward control algorithm for variation reduction, which incorporates process and product constraints. In the second part of my talk, I will show how the above mentioned methodology can be used in the determination of the optimal distribution of actuators, which are used to perform compensations. The third part of the presenetation will focused on robust design of multistage assembly processes. Specifically, I will present a fixture layout design methodology for multiple products which are assembled in a reconfigurable system. Simulations are used to prove that the aforementioned methodologies can effectively reduce variation in MMPs. Finally, I conclude my talk with future research directions.

GUEST LECTURER
Dionne Aleman: Faculty candidate

AFFILIATION
University of Florida

GUEST LECTURER
Hoang Tuy

AFFILIATION
Institute of Mathematics VAST, Hanoi

ABSTRACT
We discuss existence and stability conditions for the saddle value and the saddle point of a quasiconvex quasiconcave function depending upon a parameter. It is shown in particular that a straightforward refinement of Tuy's earlier topological minimax theorems can be obtained that yields several new minimax theorems with much weaker compactness and continuity assumptions than usual. An application to Lagrange relaxation is given that provides a rigorous foundation for decomposition methods in nonconvex global optimization.

GUEST LECTURER
Professor Qiang Huang

AFFILIATION
Department of Industrial and Management Systems Engineering, University of South Florida

ABSTRACT
Understanding fundamental phenomena is an important step to improve quality and productivity in complex systems. In traditional manufacturing, it has been observed that different error sources may result in identical error patterns on product features. This so-called error equivalence phenomenon could have dual effects on dimensional control: significantly increasing the complexity of root cause identification, and providing an opportunity to use one error source to counteract or compensate the others. In a complex system like human physiology, interaction is a fundamental mechanism of processing and coding information. Understanding complex interactions through observed physiologic signals is vital to diagnosis and treatment of a wide range of medical problems. In nanomanufacturing, relating macro/micro scale functional process variables with nanoscale nanostructure defects is a key step to implement process control for improving process yield and repeatability.

A common methodological framework of modeling and analyzing complex process/system phenomena is to integrate subject matter knowledge with mathematical modeling tools. The challenge lies in the interdisciplinary nature of the problems of interest. In this talk, the speaker will first introduce error equivalence modeling through kinematic analysis. The impact of error equivalence concept on dimensional variation control will be demonstrated through a new sequential root cause identification procedure and a novel error-compensating-error strategy. To understand physiologic interactions, a novel nonlinear dynamics model was developed to consider not only complex spatial patterns among multiple nonstationary signals, but also temporal patterns within individual signals. Validated by controlled physiologic experiments, the proposed model formulation provides better insight into interaction mechanisms, such as the interaction order, strength, and structure. In nanomanufacturing application, the speaker will discuss the research challenges and ideas for quality control of nanostructure growth processes. The ongoing activities at the speaker

GUEST LECTURER
Professor Shane Henderson

AFFILIATION
Cornell University

ABSTRACT
Joint work with Sam Steckley and Vijay Mehrotra

We investigate the presence and impact of forecast errors in the arrival rate of customers to a service system. Analysis of a large data set shows that forecast errors can be large relative to the fluctuations naturally expected in a Poisson process. We show that ignoring forecast errors typically leads to over-estimates of performance, and that forecast errors of the magnitude seen in our data set can have a practically significant impact on predictions of long-run performance. We also define short-run performance as the random percentage of calls received in a particular period that are answered in a timely fashion. We prove a central limit theorem that yields a normal-mixture approximation for its distribution for Markovian queues, and sketch an argument that shows that a normal-mixture approximation should be valid in great generality. Our results provide motivation for studying staffing strategies that are more flexible than the fixed-level staffing rules traditionally studied in the literature.

HOST/SPEAKER
Walt Ehmer, IE 1989, President and COO, Waffle House

HOST ORGANIZATION
Georgia Tech Alumni Association

DETAILS
7:00AM - 3:00PM Career fair and Conference

8:00AM - 9:00AM Morning Workshop and Breakfast - Brent Darnell, ME 1981, Owner, Brent Darnell & Associates

9:00AM - 12:30PM Career Fair Session I

12:30PM - 1:30PM Lunch Workshop - Walt Ehmer, IE 1989, President and COO, Wafle House

1:30PM - 3:30PM Career Fair Session II

3:30PM - 4:30PM Afternoon Workshop - Len Contardo, VP of Constituent Services, GT Alumni Association

3:30PM - 5:00PM Interviews - Companies will have the opportunity to invite selected alumni to their booths for informal interviews

Registration Deadline: March 21, 2007.

DATE & TIME
Wednesday, March 28, 2007 -- 7:00 AM

DURATION
All day

LOCATION
Cobb Galleria Centre

CONTACT PERSON
Brittainy Johnson at 404-894-0796

RELATED SITE
www.gtalumni.org/ccalumni

GUEST LECTURER
Youngmi Hur

AFFILIATION
Assistant Professor, Department of Applied mathematics and Statistics, Johns Hopkins University; and C.L.E. Moor Instructor, Department of Mathematics, MIT

ABSTRACT
We will start with an overview of the wavelet representation along with its applications to image processing. We will then discuss its connection with pyramidal representations such as the Laplacian, and some of existing challenges in constructing the wavelet/pyramidal representations in high dimensions.

Next, we will describe a new methodology for representing data on regular grids, which is a hybrid of the wavelet and the pyramidal representations. We will present the details of a subclass of these representations, dubbed L-CAMP. The L-CAMP methodology provides effective wavelet constructions in high dimensions in the sense that it has fast algorithms (linear complexity with small constants) for both decomposition and reconstruction, its performance (i.e., its ability to encode smoothness of an underlying data or function) is completely understood, and its localness number (measured as the sum of the volumes of the supports of the underlying mother wavelets) is extremely small.

GUEST LECTURER
Edi Schreibmann, Emory University

ABSTRACT
Image registration in medicine is necessary in order to be able to align and compare images of the same patient acquired at different points in time. The registration is essentially an optimization problem that searches for the optimal transform parameters aligning two images under the guidance of a cost function constructed from anatomical features common in the input images.

The seminar will focus on the application of image registration to improve cancer treatment using four-dimensional (4D) imaging. 4D imaging (3 spatial dimensions + time) allows continuous sampling of patient anatomy during his/her breathing cycle, improving treatment of lung cancers as the treatment plan can be customized to the observed tumor motion. In this approach, image registration is needed to identify in the 4D image, tumor position as a function of time. Our initial experience in selecting and customizing the optimization components of an image registration scheme that tracks anatomical motion in 4D images will be presented.

GUEST LECTURER
William "Bill" George

AFFILIATION
IMPACT Speaker Series

ABSTRACT
ISyE Alumnus William "Bill" George (IE 1964), professor of Management Practice, Harvard Business School and former chair and CEO of Medtronic, Inc., will be guest lecturer for the IMPACT series on April 4, 2007. The IMPACT series takes place in the LeCraw Auditorium in the College of Management building at 4:30 p.m. The series is open to the public, and reservations are not required. For more information, visit the IMPACT Speaker Series page at http://mgt.gatech.edu/fac_research/centers_initiatives/speaker.html.

George's new book, "True North: Discover Your Authentic Leadership", written with Peter Sims, is due to be released on March 15, 2007. The book includes life stories of 125 authentic leaders. "True North" also discusses how to become an authentic leader by developing yourself and being true to your beliefs and your life story in the face of many challenges and eductions that all leaders face. Included in "True North" are the moving experiences and wisdom of Starbucks' Howard Schultz, Andrea Jung of Avon Products, Dan Vasella of Novartis, GE's Jeff Immelt, Oprah Winfrey, Amgen's Kevin Sharer, Anne Mulcahy of Xerox, and many others. He has also included a series of exercises for readers, developed by George's leadership course at Harvard Business School, to help you discover your authentic leadership style. For more information, visit www.truenorthleaders.org

DATE & TIME
Saturday, April 07, 2007 -- 4:30 PM

DURATION
2 hours

LOCATION
LeCraw Auditorium, College of Management

Keynote and Plenary Speakers:
Joel Bader (John Hopkins University)
Artemis Hatzigeorgiou (University of Pennsylvania)
Hanna Margalit (Hebrew University, Israel)
Roderic Guigo (IMIM, Barcelona, Spain)
Steven Salzberg (University of Maryland)
Mario Stanke (University of G

GUEST LECTURER
Vladimir Koltchinskii

AFFILIATION
Professor, School of Mathematics, Georgia Tech

ABSTRACTWe will discuss several problems related to estimation of integral curves of a vector field based on its noisy observations. The motivation comes from applications to Diffusion Tensor Imaging, a brain imaging technique used, in particular, to recover the paths of neural fibers in the brain based on diffusion data. An estimator will be discussed for which it is possible to establish asymptotic normality (in the sense of convergence in distribution in a functional space). This provides formulas for asymptotic distributions of several statistics of interest, such as the distance from the estimated curve to a given region. [Based on a joint paper with Ludmila Sakhanenko and Songhe Cai to appear in the Annals of Statistics].

DATE & TIME
Thursday, April 19, 2007 -- 11:00 AM

DURATION
1 hour

LOCATION
Executive classroom of Main building of ISyE, GT

CONTACT PERSON
Yajun Mei

GUEST LECTURER
Marty Reiman

AFFILIATION
Bell Labs

ABSTRACT
In this talk I describe some results related to a variant of a standard inventory model. The model has a single item, a single location, two demand classes, and operates under discrete review. There is a fixed positive replenishment lead time and unlimited backlogging of unmet demand is allowed. There are no order setup costs, and both holding costs and backlogging costs are linear. If the backlogging costs are the same for both classes then this reduces to the classical model of Karlin and Scarf (1958), who showed that a 'base-stock' policy is optimal. We are interested in the case where the backlogging costs are different for the two classes. Although it is intuitively clear that the demand class with the larger backlog cost should be treated better, the optimal policy for operating this system is not known. The simplest way to give priority to one class is to satisfy all backlogs from that class before backlogs of the other class are satisfied. A stronger preference can be given by holding some items in reserve for the higher priority class. We present an asymptotic analysis, in the long lead time limit, of these priority policies. This analysis shows that the simple priority policy (without reservation) is asymptotically optimal. The priority policy with reservation is also asymptotically optimal if the reservation threshold does not grow too fast with the lead time.

DATE & TIME
Tuesday, May 15, 2007 -- 11:00 AM

DURATION
1 hour

LOCATION
Executive Classroom 228, Main

CONTACT PERSON
Jim Dai

GUEST LECTURER
Dr. Gideon Weiss

AFFILIATION
University of Haifa

ABSTRACT
Multi-class queueing networks with infinite supply of work provide a new paradigm for queueing systems under balanced heavy traffic. In particular, they allow full utilization of the servers without congestion, and the balanced heavy traffic arises naturally, and is not the result of exogenous arrivals.

In this talk we focus on the following illustrative example: We consider a multi-class queueing system with two service stations and four classes, where there are two streams of customers, one stream is processed first in station 1 and then in station 2, the other stream moves in the opposite direction. This push pull system is distinguished from the network of Rybko and Stolyar in that be assume that there is an unlimited number of customers available in each of the two streams. Hence, both service stations can be busy all the time, i.e. work at utilization rho=1. We assume processing times are memory-less. We find a family of generalized threshold policies for which this system is stable, i.e. positive recurrent, with this full utilization. We derive the steady state distribution for some of these policies.

Joint work with Anastasia Kopzon

DATE & TIME
Wednesday, May 16, 2007 -- 11:00 AM

DURATION
1 hour

LOCATION
Room 228 Executive Classroom, Main bldg

CONTACT PERSON
Alex Shapiro

GUEST LECTURER
Dr. Jianjun Shi

AFFILIATION
University of Michigan

DATE & TIME
Thursday, May 31, 2007 -- 11:00 AM

DURATION
1 hour

LOCATION
ISyE MAIN, Executive Classroom

CONTACT PERSONjharris@isye.gatech.edu

The 2007 ISyE OYSTER ROAST
DATE: Thursday, October 18, 2007
TIME: 7:00 pm - 9:00 PM
VENUE: The Trolley Barn in Inman Park, 963 Edgewood Avenue NE, Atlanta, GA 30307.

RSVP requested by October 4, 2007

To RSVP, visit www.isye.gatech.edu/oyster or call Ruth Gregory at 404-385-2627.

Dr. Bartholdi's lecture entitled "Self-organizing Logistics Systems" focuses on the complex logistical systems operated by social insects, such as ants or bees. This idea has been successfully adapted to coordinate order-pickers in a warehouse. Using a protocol called "bucket brigades," the flow of work is smoothed and bottlenecks are removed. The bucket brigade protocol has increased pick rates by 20-50% at major distribution centers.

Tuesday talks is a series of lectures sponsored by the Georgia Tech Library featuring research by Georgai Tech faculty presented for the rest of us. For more information, email prgroup@library.gatech.edu

Computational Approaches to Gene Regulation

Understanding how genes are regulated in various circumstances (e.g.,
heatshock, starvation, etc.) is a central problem in molecular biology.
The adoption of large-scale biological data generation techniques such as
the mRNA microarrays has enabled researchers to tackle the gene regulation
problem in a global way. I will survey some computational and statistical
strategies developed by our group on how to effectively use the gene
upstream sequence information in junction with mRNA expression microarray
data to dissect the gene regulatory network. I will describe in detail a
study of RacA binding activities in Bacillus Subtilis, explaining how
statistical approaches helped the biologists discover RacA's binding
sites. I will describe a new dimension reduction technique that has been
applied successfully to our gene regulation studies and show a cute
theorem supporting the technique.

Host: Xiaoming Huo & Jeff Wu

Traditionally companies have focused on improving their own internal business processes when faced with pressures to operate more efficiently. However, a system-wide focus, e.g., a collaborative focus, opens up cost saving opportunities that are impossible to achieve with an internal company focus. With the availability of timely and accurate information, new collaborative opportunities which create mutual benefit by taking advantage of operational synergies between buyers, sellers, or buyers and sellers are now arising. Collaborative logistics is viewed by many logistics professionals as a promising opportunity for reducing logistics costs and therefore increasing profitability and economic prosperity. Hence there is a need to develop a fundamental knowledge base for designing, analyzing, implementing, and managing such collaborative logistics initiatives.

The workshop will focus on efforts in that direction and will have a variety of talks on topics related to models and methods of collaboration, including supply chain coordination, auctions, computation and games, learning and games, and logistics collaboration in practice. Furthermore, an overview of collaborative logistics research at Georgia Tech will be presented. A 15-minute discussion after each talk, a lunch break, and two coffee breaks during the day will provide ample opportunities for
interaction among the participants. The speakers include:
* Michael Ball, University of Maryland
* Jun-Sheng Li, Transplace
* David Parkes, Harvard University
* Andreas Schulz, MIT
* Vijay Vazirani, Georgia Tech
* Rakesh Vohra, Northwestern University
To reserve your place in the workshop please respond via e-mail to oergun@isye.gatech.edu by September 12, 2007.

Related Documents
Collaborative Logistics Workshop Abstracts
Collaborative Logistics Workshop Agenda

Title: Mingling: Mixed-Integer Rounding with Bounds

As in previous occasions the seminars will include both formal presentations and informal discussion about open problems or current work by students. However only formal presentations will be announced in the usual ISyE lists, so if you would like to subscribe to the seminar's email list please visit http://www2.isye.gatech.edu/ mailman/listinfo/dos_list. If you would like to be a speaker you can let us know in person or by email. You can also visit the seminar's web page at http://www2.isye.gatech.edu/dos.

Speaker: Tirthankar Dasgupta, Georgia Tech

Abstract: In this paper, we discuss an integrated approach for estimation and reduction of measurement variation (and its components) through a single parameter design experiment. The noise factors are classified into a few distinct categories based on their impact on the measurement system. A random coefficients model that accounts for the effect of control factors and each category of noise factors on the signal-response relationship is proposed. A suitable performance measure is developed using this general model, and conditions under which it reduces to the usual dynamic signal-to-noise (SN) ratio are discussed. Two different data analysis strategies -- response function modeling (RFM) and performance measure modeling (PMM) -- for modeling and optimization are proposed and compared.

Tirthankar Dasgupta was a PhD student in ISyE working with Professor Jeff Wu. He is now a postdoc and will join Department of Statistics at Harvard University in the next year.

For Additional information about the ISyE Statisitcs Group, please visit: http://www2.isye.gatech.edu/statistics/seminar_series.php

Title: An Optimization Approach for Scheduling Wine Grape Harvest Operations

Abstract: This presentation discusses a practical tool for optimally scheduling wine grape harvesting operations taking into account both operational costs and grape quality. We solve a mixed-integer linear programming model to support harvest scheduling, labor allocation, and routing decisions. A quality loss function is used to represent wine quality reduction at each vineyard block due to premature or deferred harvest with respect to an optimal date. We present computational results which show that the proposed tool could be used to support grape harvest planning in a large vineyard, at both a tactical and operational level.

Pizza and refreshments sponsored by The Supply Chain & Logistics Institute (SCL).

Speaker: Demetrio Lagana
Assistant Professor of Operations Research at University of Calabria, Italy

Title: The Capacitated Arc Routing Problem

Abstract:

The Capacitated Arc Routing Problem (CARP) consists of determining a set of least cost capacitated vehicle routes servicing a set of arcs. Applications of the CARP arise in garbage collection, snow removal, street sweeping and gritting, mail delivery, meter reading, school bus routing, etc.

While the CARP is a central problem in arc routing and has been known for a long time, it is still almost exclusively tackled by means of heuristics. The only available exact method for the CARP is a parallel branch-and-bound algorithm proposed by Hirabayashi et al. (1992), and Kiuchi et al. (1995), in which a lower bound is computed at each node through a node duplication lower bounding procedure (Hirabayashi et al., 1992). Transformations of the CARP into an equivalent node routing problem (namely the Capacitated Vehicle Routing Problem, CVRP) have been proposed by Pearn (1987), Longo et al. (2006), and Baldacci and Maniezzo (2006). Lower bounds have been obtained by Belenguer and Benavent (1992, 1998, 2003) through cutting plane procedures, and Wels (1994) has proposed valid inequalities and separation procedures for the directed version of the CARP.

Recently, Ghiani et al. (2007) proposed a pure binary linear integer program for the undirected CARP, on the basis of an extension of a previous formulation by Belenguer and Benavent (1998), and they developed a fully automated branch-and-cut algorithm. Computational results indicate that the algorithm is able to solve, for the first time ever, all the benchmark instances proposed by DeArmon (1981), and Benavent et al. (1992).

Pizza and refreshments sponsored by The Supply Chain Logistics Institute (SCL).

Title: Approximation of Haar Distributed Matrices and Limiting Distributions of Eigenvalues of Jacobi Ensembles

Abstract: I will first present tools to approximate the entries of a large dimensional real and complex Jacobi ensembles with independent complex Gaussian random variables. Based on this, we obtain the Tracy-Widom law of the largest singular values of the Jacobi emsemble. Moreover, the circular law, the Marchenko-Pastur law, the central limit theorem, and the laws of large numbers for the spectral norms are also obtained.

Hosted by: Xiaoming Huo

Title: Examples of Industrial Problems Leading to Statistical Research

Abstract: Many of the problems that are faced in industry can be solved with methods that have already been developed. Occasionally there are new problems that are faced that require original research. For a research statistician working in industry, these types of problems are a joy to encounter. In this talk I will discuss several examples of problems faced at Procter & Gamble that have led to subsequent statistical research.

Host: Roshan J. Vengazhiyil (Contact Roshan if you want to speak to the visitor.)

Title: The Challenges and Benefits of the Humanitarian Supply Chain vis-a-vis the Private Sector Supply Chain

Abstract: This is an introductory presentation from the eyes of a practitioner who has been privileged to sit on both sides of the fence. It looks at some of the challenges faced and benefits found in supply chains for both humanitarian organizations and private sector organizations. It also highlights some of the differences in the drivers of both demand and supply in both types of organizations. The presentation will cover the challenges presented by the environment (political, social, economical, and technological) in which humanitarian supply chains function, including an overview of challenges presented by collaboration with partners and the issue of ethics during disaster relief.

Pizza and refreshments sponsored by The Supply Chain Logistics Institute (SCL).

Abstract: We are creatures of habit. When the usual state of affairs is no longer, we want to be aware of the change. (Is there global warming? Has the incidence rate of breast cancer increased? Is there a change in the volatility of the stock market?) Unfortunately, often there is no one around to tell us that things are not what they used to be, and we have to rely on a sequence of (invariably noisy) observations to discern that things are different. Intuitively, if the behavior of "recent" observations seems "abnormal", we would be prone to declare that a change is in effect. As usual, the problem is in the details: what do "recent" and "abnormal" mean in practice? Moreover, if we are trigger-happy, there will be many false alarms; if we are too hesitant, the delay between a real change and its detection may be very costly. Subject to a constraint on the rate of false alarms, what is the best average delay (between a true change and its detection) attainable? What is a good procedure?

This talk will present a short review of the change point problem and some of its solutions. The early proposed methods turned out to be optimal in the ideal case that both pre-change and post-change distributions (of an independent sequence of observations, iid pre-change and iid post-change) are known. More recent research relaxed these assumptions. The talk will concentrate on a set of scenarios where progressively less is known about the distributions of the observations, culminating in nonparametric procedures for cases where nothing (other than independence) is assumed about the pre- and post- change distributions.

Please contact Yajun Mei (yajun.mei@isye.gatech.edu) if you want to talk to the speaker.

Title: Polynomial Spline Confidence Bands for Regression Curves

Abstract: Asymptotically exact and conservative confidence bands are obtained for a nonparametric re

Title: Syndromic Surveillance: Improving Public Health Systems Preparedness through Advances in Statistical Modeling

Abstract: Syndromic Surveillance has emerged as a new public health surveillance paradigm to identify disease clusters earlier than would conventional reporting of confirmed cases. The approaches used in syndromic surveillance are similar to those in traditional statistical monitoring. We develop the multivariate cuscore (MCS) control chart and the high density (HD) control chart to aid in early detection. Existing relevant approaches in the literature include nonparametric analogues and the multivariate changepoint (MCP). We demonstrate how these statistical methods can be applied on influenza-like illnesses (ILI) data from the US and France. Challenges are presented by the fact that ILI data are non-normal, not independent, and often nonstationary. In addition, we explore how these statistical methods can be refined to take advantage of prior information about health system characteristics such as increases in over-the-counter medication sales and seasonal patterns of outbreaks.

Biosketch: Harriet Black Nembhard is an associate professor of industrial and manufacturing engineering at the Pennsylvania State University and director of the Laboratory for Quality Engineering and System Transitions (QUEST). Dr. Nembhard

Abstract:
Treatment planning for radiation therapy is inherently complex due to the number of input parameters involved. The complexity is amplified by the uncertainty of target shapes due to organ motion, by dose estimation, by availability of biological information, and by competing multiple clinical objectives within the planning procedure. In this talk, we describe some of our approaches to cancer treatment design pertaining to these issues. Various optimization methods will be contrasted, and computational challenges will be discussed.

Time and Place:
11am-12 Oct. 19, 2007. Rm. 109, Instructional Center

Speaker:
Dr. Youxun Shen, CFA
VP, Goldman Sachs

Abstract
In this talk, a literature review of algorithmic (algo) trading will be provided. Dr. Shen will also explain why and how quantitative techniques could play a vital role in algo trading.

The need to solve a large number of maximum flow problems as part of a greater computational procedure appears in literature on game theory, integer programming and stochastic programming. However, simply using a black-box maximum flow solver can often lead to an exorbitant amount of unnecessary computations. In this talk, we discuss how to rapidly solve an online sequence of similar maximum flow problems using efficient reoptimization techniques. We will then extend our work to efficiently solve a robust minimum cut problem.

This is joint work with Dr. Ozlem Ergun, assistant professor in the Stewart School of ISyE.

Speaker:
Brian Lewis, Ph.D. IE 2005
Vice President, Professional Services
Vanguard Software Corporation
Cary, NC 27518

We will present an overview of the Vanguard SytemTM software solution and how faculty can apply it to their courses and research.

The Vanguard SystemTM is a comprehensive planning and analysis solution for improving the quality, reliability, and speed of management decisions. It does this by helping businesses:

Southern Company uses coal to generate much of the electricity that we all use, and John Harris of that company will speak on their use of Operations Research techniques to improve use of the railcars that are used to move it around the south.

The meeting is open to all interested parties, and is free of charge. Refreshments will be served, and there will also be time to network with fellow OR professionals in the Atlanta community. Please pass this email along to any associates whom you think might be interested.

The meeting will be in the Executive Classroom (Room 228) of the ISyE Main Entrance Bldg. (755 Ferst Drive, formerly the Dupree School of Management Building). (Same location as previous recent meetings.)

Here's a link to a map of the campus: http://www.isye.gatech.edu/visitors/maps/

There is plenty of (no longer free) parking next to the ISyE building.

We look forward to seeing you there. Please contact me if you have questions or suggestions, or if you wish to be added to / removed from the mailing list for future meetings.

Speaker: John Guerard (McKinley Capital)

Abstract: Mr. Guerard will present an updated version of a paper presented in New York in June at the International Symposium on Forecasting (ISF). The paper illustrates the McKinley Capital Management implementation of quantitative techniques to portfolio construction in Emerging Markets and the World universes.

Speaker Bio:
John B. Guerard, Jr., Ph.D., is Director of Quantitative Research at McKinley Capital Management, in Anchorage, Alaska. He earned his AB in Economics from Duke University, MA in Economics from the University of Virginia, MSIM from the Georgia Institute of Technology, and Ph.D. in Finance from the University of Texas, Austin. John taught at the McIntire School of Commerce, the University of Virginia, and Lehigh University. He worked with the DAIS Group at Drexel, Burnham, Lambert, Daiwa Securities Trust Company, Vantage Global Advisors, and served on the Virtual Research team at GlobeFlex Capital. John co-managed a Japanese equity portfolio with Harry Markowitz at Daiwa Securities Trust Company. While serving as Director of Quantitative Research at Vantage Global Advisors (formerly MPT Associates), Mr. Guerard was awarded the first Moskowitz Prize for research in socially responsible investing. Mr. Guerard has published several monographs, including The Handbook of Financial Modeling (Probus, 1989, with H.T. Vaught), Corporate Financial Policy and R&D Management (Wiley, 2006, second edition), and Quantitative Corporate Finance (Springer, 2007, with Eli Schwartz). John serves an Associate Editor of the Journal of Investing and The International Journal of Forecasting. Mr. Guerard has published research in The International Journal of Forecasting, Management Science, the Journal of Forecasting, Journal of Investing, Research in Finance, Research Policy, and the Journal of the Operational Research Society.

====================================================

TITLE: Random Search Methods for Simulation Optimization

ABSTRACT: We will discuss the use of random search methods for optimizing complex systems via simulation. We will consider both discrete and continuous parameter problems. Our focus will be on averaging methods for discrete
simulation optimization and on resampling methods for continuous simulation optimization. We will present asymptotic convergence results and also numerical results.

Joint work with Andrei A. Prudius

Title: Modeling and Analyzing High-Frequency Financial Data

Abstract: Volatilities of asset returns are central to the theory and practice of asset pricing, portfolio allocation, and risk management. In financial economics, there is extensive research on modeling and forecasting volatility up to the daily level based on Black-Scholes, diffusion, GARCH, stochastic volatility models and implied volatilities from option prices. Nowadays, thanks to technological innovations, high-frequency financial data are available for a host of different financial instruments on markets of all locations and at scales like individual bids to buy and sell, and the full distribution of such bids.

The availability of high-frequency data stimulates an upsurge interest in statistical research on better estimation of volatility. This talk will start with a review on low-frequency financial time series and high-frequency financial data. Then I will introduce popular realized volatility computed from high-frequency financial data and present my work on wavelet methods for analyzing jump and volatility variations and the matrix factor model for handling large size volatility matrices. The proposed wavelet based methodology can cope with both jumps in the price and market microstructure noise in the data, and estimate both volatility and jump variations from the noisy data. The matrix factor model is proposed to produce good estimators of large size volatility matrices by attacking non-synchronized problem in high-frequency price data and reducing the huge dimension (or size) of volatility matrices. Parts of my talk are based on joint work with Jianqing Fan, Qiwei Yao, and Pengfei Li.

Accelerated Recurrence Time Models

For the analysis with recurrent events, we propose a generalization of the accelerated failure time model to allow for evolving covariate effects. These so-called accelerated recurrence time models postulate that time to expected recurrence frequency, upon transformation, is a linear function of covariates with frequency-dependent coefficients. This modeling strategy shares the same spirit as quantile regression. An estimation and inference procedure is developed by generalizing the celebrated Powell's (1984, 1986) estimator for censored quantile regression. Consistency and asymptotic normality of the proposed estimator are established. An algorithm is devised to attain good computational efficiency. Simulations demonstrate that this proposal performs well under practical settings. This methodology is illustrated in an application to the well-known bladder cancer study.

This talk is based on joint work with Limin Peng.

We will meet for a reception (with refreshments) at 5:30 PM Thursday November 29, 2007 in the Executive Classroom at the ISyE School at Georgia Tech. John

Title: Multi-Period Remanufacturing Planning with Uncertain Quality of Inputs

Abstract: In this paper we consider production planning of remanufactured products when inputs have different and uncertain quality levels, and there are capacity constraints. This situation is typical of most remanufacturing environments, where inputs are product returns (also called cores). Production (remanufacturing) cost increases as the quality level decreases, and any unused cores may be salvaged at a value that increases with their quality level. Decision variables include, for each period and under a certain probabilistic scenario, the amount of cores to grade, the amount to remanufacture for each quality level and the amount of inventory to carry over for future periods for un-graded cores, graded cores, and finished remanufactured products. Our model is grounded with data collected at a major OEM that also remanufactures. We formulate the problem as a stochastic program, and illustrate how the deterministic version of the problem yields solutions that cannot be implemented in practice. The stochastic program, although a large linear program, can be solved easily using Cplex. We provide a numeric study to generate insights into the nature of the solution.

Pizza and refreshments sponsored by The Supply Chain Logistics Institute (SCL).

"Ecomagination puts into practice GE's belief that financial and environmental performance can work together to drive company growth, while taking on some of the world's biggest challenges. Learn about the GE commitment to products and services that are as economically advantageous as they are ecologically sound."

http://ge.ecomagination.com

The lecture is part of ISYE 8813B/Q, Industrial Ecology and Natural Systems, and will be in IC 105. The lecture is open to the entire campus community.