U-M to be part of NSF high-performance computing initiative
ANN ARBOR—The University of Michigan has been selected by the National Science Foundation to help create the National Partnership for Advanced Computational Infrastructure (NPACI) to provide unprecedented computing power for advanced scientific and engineering research.
The University of California-San Diego (UCSD) is the lead institution responsible for directing and managing the development of NPACI. The U-M is one of 37 research universities, federal laboratories and private companies that will provide computing facilities and technical expertise for NPACI’s national computational infrastructure.
NPACI will give scientists the high-performance computers, high-speed networks, applications software and technical expertise they need to solve complex, data-intensive and computer- intensive problems such as modeling the effects of automobile crashes, predicting global climate change, improving medical imaging and diagnostic technology, and tracking the flow of subsurface pollution.
On March 27, the National Science Foundation (NSF) agreed to begin negotiations with the University of California-San Diego funding package to establish NPACI.
“Not only will NPACI have high-performance computers, we will have the capability to link them together with very high- speed networks to solve complex problems involving massive data sets,” said William R. Martin, a U-M professor of nuclear engineering and radiological sciences, who is project director for the U-M NPACI initiative and will serve on NPACI’s executive committee. “Our first task will be learning how to coordinate the processing power of many computers, perhaps from different vendors, and get them all working simultaneously to solve a single application.”
According to Martin, approximately $800,000 in NSF funding will be earmarked for annual support of parallel computing facilities at U-M and infrastructure support for research programs in computational science and engineering. The U-M and the U-M College of Engineering will contribute $800,000 in matching funds.
Faculty who will play a major role in the U-M’s involvement with NPACI include Edward S. Davidson, professor of electrical engineering and computer science; Randall L. Frank, director of information technology for the Computer Aided Engineering Network; Richard C. Rockwell, research scientist at the U-M’s Institute for Social Research, and Quentin F. Stout, professor of electrical engineering and computer science.
“The U-M has many strengths that make it a valuable addition to this partnership,” said Martin. “The U-M’s Center for Parallel Computing has the facilities, staff and expertise to handle computer-intensive research. We have a proven track record in distance learning and Web-based learning, which will be an important component of NPACI. Plus, the U-M is well-known for its high-quality research program in the social sciences—an important new applications area for high-performance computing.”
“Social scientists work with massive data sets made up of hundreds or thousands of interdependent variables,” said Rockwell. “With advanced computing capability and new software, NPACI will make it possible for us to visualize or model complex statistical relationships between multiple variables in ways we cannot even attempt with current computer technology.”
NPACI-funded computing resources and research initiatives at U-M will include:
—An IBM SP2 multicomputer and technical user support for the nationwide NPACI research community.
—Integration of the U-M Digital Library’s storage architecture and data management system into NPACI.
—Distance learning technology and Web-based instructional materials to train NPACI users.
—Summer training workshops on high-performance computing technology for social science researchers.
—Infrastructure support for computational science and engineering research programs. Initial areas targeted for support will be:
—An engineering applications simulation of the impact of vehicle collisions.
—Simulation of subsurface, multi-phase flow involved in petroleum reservoirs and pollution remediation.
—Software for large-scale microstructural analysis of composite materials.
A second partnership—the National Computational Science Alliance (NCSA), to be directed by the University of Illinois, Urbana-Champaign—also was approved by NSF on March 27.