Media Contact: Denine Hagen, (858) 534-2920

UCSD RECEIVES $4.5M FROM NEW FEDERAL PROGRAM
TO PROMOTE INNOVATION IN INFORMATION TECHNOLOGY

Projects will maintain U.S. leadership in computer research

The National Science Foundation (NSF) today announced its first grants under the $90 million Information Technology Research program (http://www.itr.nsf.gov/). Five University of California, San Diego (UCSD) projects, selected from a nationwide pool of 1,400 proposals, have been awarded $4.5 million. The NSF made a total of 210 awards.

“Information technology is undoubtedly revolutionizing every aspect of our economy, and we congratulate the NSF for making this kind of investment in long-term research,” said Robert W. Conn, dean of the UCSD Jacobs School of Engineering. Conn says UCSD research funded through the project will advance software, computer architectures, and networked computing for academic research, and will eventually lead to innovations in the commercial marketplace.

One of the leading projects funded through the Information Technology Research (ITR) program is research on large-scale “steerable” software simulations across networks to model the form and function of cells. This $2.5 million project, entitled “Virtual Instruments: Scalable Software Instruments for the Grid” is led by Francine Berman, professor of computer science and engineering at the Jacobs School of Engineering and a fellow of the San Diego Supercomputer Center.

The software will simulate critical cellular behaviors over a huge set of possible environmental parameters. Researchers will be able to steer the computation to target particular areas of the simulation based on partial results that evolve continually over the life of the simulation. Initially, prototyping will be done using simulations focused on synaptic transmission in the brain and between nerve and muscle cells. The software will be developed by a team of computer scientists at UCSD and University of Tennessee and disciplinary scientists at UCSD, Salk Institute and Carnegie Mellon University.

Additional UCSD projects receiving ITR grants include:

$625,000 to UCSD (Total Project funding $11.8 million) “The GriPhyN Project: Towards Petascale Virtual-Data Grids” led by the Universities of Florida and Chicago

The Grid Physics Network, or GriPhyN, will lay the groundwork for a computer data grid of unprecedented speed and power. GriPhyN initially aims at giving scientists a tool to interpret the vast amounts of data expected to flow from the world’s most ambitious physics and astronomy experiments, but it could also have applications in the business world and elsewhere. At UCSD, Reagan Moore of the San Diego Supercomputer Center and Keith Marzullo, professor of computer science and engineering at the Jacobs School of Engineering, will work to integrate UCSD-developed technologies into this grid environment.

$447,682 “Algorithms for Machine Perception based on Visual Cortex Models” led by Irina Gorodnitksy, professor of cognitive science.

Even the simplest of animals, such as insects, have remarkable perception capabilities, which are difficult for present-day engineered systems to duplicate. This project seeks to understand how the brain accurately organizes and processes visual information and to develop machine vision systems that mimic brain visual function. The novel aspects of this project include the innovative use of scalp electrodes to probe macro-scale processes of the visual cortex, investigation of encoding of sequences of patterns rather than single presentations, use of innovative mathematical models, and recognition of the existence of several different, task modulated, low-level perceptual mechanisms. The project is an international collaboration between laboratories at UCSD, the Algerian Ecole Nationale Polytechnique, and the Brain Science Institute in Japan.

$390,404 Application-Specific Reconfigurable Microarchitectural Enhancements for Embedded Processors in High-Performance Hardware/Software Codesigns” led by Alex Orailoglu, professor of computer science and engineering, Jacobs School of Engineering.

As more and more electronic solutions arise such as networking, wireless communications, and visual imaging, there is a need for customizable, reconfigurable imbedded computer processors to accommodate the various niches. The challenge is to keep the volume benefits of generic, general purpose processors but create an adaptable processor which is suitable to a variety of applications. Orailoglu will attempt to lay the foundation for a new type of imbedded computer processor architecture that is cost effective while providing flexibility to various devices and applications.

$344,691 “Validating Simulation To Observed Data With Source Coding Methods” led by Matthew Kennel, physicist, Institute for Nonlinear Science.

Kennel’s project will investigate statistical techniques and algorithms to improve the match of computer simulations with experimental observations complicated by “deterministic chaos.” In many modern applications of Newton’s theory of gravitation, for example, the presence of deterministic chaos means that the trajectories of an object in space never repeat, making it difficult to match computer simulations with observations. The results from this project should improve that match.

$168,498 “Innovative Software for Large-Scale Nonlinear Optimization” led by Philip Gill, professor of mathematics.

New methods and advanced software will be developed for solving the large-scale constrained optimization problem, which may be broadly defined as the problem of finding the “best” or “optimal” strategy from among many competing strategies—some of which may be subject to restrictions or constraints. Constrained optimization has wide application in manufacturing, engineering and science, ranging from the optimal control of spacecraft to the design of an entry in the America’s Cup yacht race.