The National Science Foundation (NSF) has awarded Cornell University $1.5 million for a new facility for research on multiscale problems in materials science and molecular biology.

The award, from the NSF's directorate for Computer and Information Sciences and Engineering, will support a project involving four universities and located at the Cornell Theory Center (CTC), Cornell's high-performance computing center. The goal of the so-called multiscale research project is to develop the scalable hardware and software systems needed to simulate critical technical and scientific problems, from analyzing a cracking piston to determining the way in which a protein folds.

Collaborating institutions include Cornell, Iowa State University, the University of Notre Dame and the University of Delaware. Teams of computer scientists and visualization researchers from these institutions will work on the project.

Multiscale research requires computing methods that can make the transition between microns and meters. It also requires tools that can translate between methods developed for different disciplines based on various theoretical approaches. In addition, methods first developed for small exploratory systems and running on desktop computers often need to be scaled up to parallel supercomputers in order to achieve the sought-for results.

"We are charged with developing the simulation tools of the 21st century," says principal investigator Keshav Pingali, professor of computer science at Cornell. He will oversee development of parallel software that will be used on cluster computing systems at Cornell, the University of Delaware and Notre Dame. This work is being performed at Cornell with the Cornell Fracture Group in the Department of Civil Engineering and the Computational Biology Group in the Department of Computer Science under NSF funding totaling more than $5 million.

The new facility will function at two levels, with the lower tier providing low-cost computation and visualization for desktop computers for integration with the upper tier of high-power computers. This will make possible large production runs and high-end visualization. To this end, researchers will seek methods to integrate new, off-the-shelf desktop computers with advanced resources, such as CTC's AC3 Velocity parallel cluster computer and the virtual reality laboratories at CTC and Iowa State.

The research teams face challenges in developing computer programming and architecture design, as well as creating intuitive and simple user interfaces. This work will support current programs in multiscale research by helping to develop new research tools. Such tools might link atomic-level information about subtle shifts in patterns in a crystal or might make it possible to adjust the simulated structure of a potential drug as it floats near a target enzyme.

"NSF support through computational infrastructure funding underscores the importance of providing scalable low-cost tools to meet the challenges of multiscale research," says CTC Director Thomas F. Coleman. The project, he notes, will allow CTC researchers to work with a broader research community using standard desktop computers.

CTC is a high-performance computing and interdisciplinary research center focused on making parallel computing a usable tool for computational science and engineering. Since its founding in 1984, it has worked with a variety of hardware and software companies in research and development projects. CTC receives funding from Cornell, New York state and a number of federal agencies and corporations.