Special Acknowledgement to the National Science Foundation (DMI-0537840)
Bridging the Achievement Gap with OR
Lincoln J. Chandler
Operations Research Center
Massachusetts Institute of Technology
My research concerns the application of statistical analysis and operations research to schools and school systems. The setting for my current work is an elementary school district in which the administration has perceived a significant difference in the academic performance of its students, particularly when grouped by race. This phenomenon is commonly referred to the Achievement Gap and its existence has been documented in various types of educational settings. There is considerable interest in closing the Achievement Gap; fortunately, the district with which I am working has unusually rich longitudinal data on a host of relevant variables. Using these data, I seek to perform analyses aimed at characterizing the gap, identifying the factors that seem strongly associated with it, and uncovering the reform strategies that might be most effective in closing the gap.
Scheduling an Air Taxi Service: A Local Search Based Approach
School of Industrial and Systems Engineering
Georgia Institute of Technology
Several companies today are attempting to offer consumers an on-demand air taxi service in which the traveler calls one day, or a few days in advance, to schedule transportation. DayJet Inc. is a startup company in this domain that will soon provide this service. They plan to use smaller aircraft to meet the customized needs of the traveling public for greater flexibility in scheduling and access to almost every airport in the country. Since fixed flight schedules will not be used, a key challenge in setting up this service is the development of a scheduling engine that takes requests for transportation and schedules planes and pilots in a cost effective way to satisfy these requests. A dispatcher has to decide which planes and pilots to use to satisfy the reservations and what the plane and pilot itineraries will be. This problem can be solved optimally for relatively small problem sizes of nine or less, but practically, the problem must be solved for hundreds and even thousands of planes. A parallelized local search scheme using mixed integer programming has been implemented to solve the large scale problems and has provided excellent results.
On Properties of the Hohmann Transfer
Mechanical Engineering and Material Science
In this work, we present a complete study of the Hohmann transfer maneuver between two circular coplanar orbits. After revisiting its known properties, we present a number of supplementary properties which are essential to the qualitative understanding of the maneuver. Specifically, along a Hohmann transfer trajectory, there exists a point where the path inclination is maximum: this point occurs at midradius and is such that the spacecraft velocity equals the local circular velocity. This implies that, in a Hohmann transfer, the spacecraft velocity is equal to the local circular velocity three times: before departure, at midradius, and after arrival. In turn, this allows the subdivision of the Hohmann transfer trajectory into a region where the velocity is subcircular and a region where the velocity is supercircular, with the transition from one region to another occurring at midradius.
Also, we present a simple analytical proof of the optimality of the Hohmann transfer and complement it with a numerical study via the sequential gradient-restoration algorithm. Finally, as an application, we present a numerical study of the transfer of a spacecraft from the Earth orbit around the Sun to another planetary orbit around the Sun for both the case of an ascending transfer (orbits of Mars, Jupiter, Saturn, Uranus, Neptune, and Pluto) and the case of a descending transfer (orbits of Mercury and Venus)
Optimal Control of a Make-to-Stock System with Adjustable Production Rate
Maria E. Mayorga
Industrial Engineering and Operations Research
University of California at Berkeley
We consider a multi-class make-to-stock system served by a single server with adjustable production rate. At each order arrival and production completion time the decision maker must determine production rate and choose a production decision (i.e., whether to produce an item to stock or satisfy a backorder) and a rationing decision (i.e., whether to satisfy a new order from stock or backorder). In this paper we characterize the structure of optimal capacity adjustment, production, and stock rationing policy for both finite and infinite horizon problems. We show that optimal policy is monotone in current inventory and backorder levels, and characterize its properties. In a numerical study we compare the optimal policy with heuristic policies and show that the savings from using optimal policy are significant.
An Investigation on the Impact of Transactive Memory Systems on Virtual Team Performance
Industrial and Operations Engineering
University of Michigan
The area of knowledge management and virtual organizations continues to receive considerable attention in business and research. Within this arena, studies have explored a variety of processes for knowledge transfer in organizational group settings. One of the mechanisms that has been identified as an emerging research topic is a transactive memory system. Essentially, these systems represent formations of collective memory within small groups. Another topic that has recently received significant attention is the idea of virtual organizations and teams. Studies indicate that despite the increasing interest in virtual organizations, there have been few empirical studies performed in this area. While most transactive memory systems studies have traditionally addressed its static nature, only recently have studies explored the evolutionary, rather than the static, nature of transactive memory systems; however, the evolutionary and static perspectives are limited in their applications and benefits to virtual environments. The very nature of transactive memory constrains the ability to utilize this mechanism in virtual organizations. My dissertation project intends to empirically determine how transactive memory systems can emerge and thrive in a virtual environment, despite these constraints. Suggested observations include determining, through communication practices and knowledge repositories (among other means), whether a transactive memory system can be created in a virtual environment, and which conditions favor or hinder this creation. Additionally, the study will investigate which transactive memory practices result in improved virtual group performance.
A Methodology for Prediction and Analysis of Chemotherapy Response in High Risk AML Patients Using Temporal Gene Expression Profile
Industrial and Management Systems Engineering
University of South Florida
Low rates of success of Acute Myeloid Leukemia (AML) chemotherapy regimens and high treatment related mortality have made proper prediction of chemotherapy response a critical translational research issue. The primary aims of this research are 1) to develop a comprehensive methodology for extracting biomarker patterns for response prediction based on gene expression data from peripheral blood and bone marrow samples collected before chemotherapy, and 2) to develop an approach to analyze functional impact of the biomarker genes on treatment response using temporal gene expression profile from samples collected before, during, and after chemotherapy. Samples will be procured from high-risk AML patients enrolled in a Moffitt Cancer Center clinical trial that will consist of treatment with Daunorubicin and Cytarabine followed by intravenous Busulfan as conditioning treatment for autologous peripheral blood stem cell transplantation.
The prediction methodology would consist of: wavelet denoising of gene chip image, a series of data mining steps to obtain a minimal robust set of genes, generation of effective gene patterns using combinatorial optimization, and finally the construction of a discriminant for bedside decision making. The interdisciplinary team has significant federal and corporate funded research experience in modeling and optimization, cancer biology, biostatistics, and clinical AML research.
Dynamic Pricing in an Order Driven Market
Damon P. Williams
Industrial and Operations Engineering
University of Michigan
Consider an investor in an order driven market trading a single commodity. Periodic demand is seen for both purchases and sales based on the prices set by the investor. That is to say, in each period the investor sets a buy and sell price. Demand for both buys and sales are modeled as additive independent (random) functions of the price set. Holding costs are accrued for holding (or backordering) the commodity in inventory. The objective is to minimize the total discounted expected costs yielding a dynamic pricing policy for the investor. Applications include a simple limit order book and e-commerce transactions for several large online retailers.