PI: Candace Arai Yano
Co-PI:Robert C. Leachman
The Department of Industrial Engineering and Operations Research (IEOR) has 11 faculty, 80 graduate students, and over 100 undergraduates. The department offers courses and performs research in the fundamental methods of operations research (optimization, probability models, statistics, and decision theory) as well as the application of these methods in areas such as quality and yield improvement, production planning and scheduling, reliability models, facilities design, and automated assembly. In the most recent National Research Council survey, the department was ranked second nationwide among industrial engineering departments.
Two major areas of research within the department are: (1) production planning and scheduling for electronics manufacturing (from semiconductor fabrication to assembly of finished goods) and (2) yield management and quality improvement, particularly in semiconductor fabrication. Several faculty in the department are heavily involved in the development of state-of the-art analytical and computer-based tools that provide decision support related to these issues. Indeed, several of the faculty have had long-standing research relationships with the microelectronics industry, either individually, or collectively through such programs as the Sloan Foundation Competitive Semiconductor Manufacturing program and through consortia such as SEMATECH. Both of the research projects described above entail the development of complex optimization algorithms operating on very large databases and/or the development of large-scale Monte-Carlo computer simulations. Many aspects of these computations could be implemented in a parallel mode, and indeed, would provide better results more efficiently in this way.
These areas of research pertain to large and rapidly growing industry segments. Sales in the U.S. consumer electronics industry alone now exceed $68 billion per year, and annual sales of the U.S. semiconductor industry exceed $30 billion. In every phase of electronics manufacturing, firms are concerned with improving speed of delivery, reducing inventory, and improving quality. Not only do the faculty in the department perform research along these lines, but we also teach courses that enable our graduates to tackle related problems. A significant portion of our graduates go on to work in the microelectronics industry or for organizations that provide direct consultative or technical support to the microelectronics industry. They are often recruited to develop CPU- and data-intensive applications such as large scale production scheduling systems, or to develop high-resolution computer simulation and animation tools to support engineering design and analysis.
The Department of Industrial Engineering and Operations Research lacks the infrastructure to support parallel approaches to optimization and simulation. At the moment, the best available alternative has been multi-tasking on a single, fast Sun Workstation. (Although we have a network of Sun Workstations for use by faculty and graduate students, the machines are of such widely varying vintages that it is impractical to use the network for truly parallel computation.) Within the past five years, there has been significant activity in the development of new parallel methods for a variety of mathematical optimization problems computer simulation methods for which non-parallel methods already exist. For some types of problems, the ability to parallelize certain computations has led to fundamentally new ways of thinking about their solution. Access to a network of Intel processors would allow us to move quickly toward making advances in this direction.
February 1999