RELEASED: 5/28/97

CHAMPAIGN, Ill. - Using the SGI/Cray Origin2000 supercomputer at the National Center for Supercomputing Applications (NCSA) at the University of Illinois at Urbana-Champaign, a group of University of Illinois scientists set a new world record in solving aircraft scatterer problems of unprecedented size.

Earlier this month, scientists, part of the Center for Computational Electromagnetics, Department of Electrical and Computer Engineering, UIUC, solved an equation involving more than 1.9 million unknown variables.

"This solution is the first of its kind ever in the field of radar cross-section calculation," said Professor Weng Chew, director of the Center.

The radar cross-section of an aircraft is a measure of how "visible" an aircraft is to radar detectors. The calculation can be used to identify aircraft or to give them "stealth" capabilities, making the aircraft invisible when used with radar absorption materials.

Until recently, the solution to that equation for frequencies above 1 gigahertz (GHz) microwave frequency has eluded scientists due to the large number of unknowns, which increases proportionally with the size of the aircraft and the frequency of the rada r beam. Previous methods for solving the scatterer problem were impractical because of the enormous amount of supercomputer time needed to solve them. For example, a calculation with 1 million unknown variables would take almost 2,000 hours of computer ti me.

Working with principal Research Scientists, Dr. Jiming Song and Dr. Caicheng Lu, Chew has developed a very efficient algorithm for calculating the scatterer solution of very large objects. Using this algorithm, solving an equation with 1 million unknown v ariables would take only 5 hours of computing time. The algorithm is based on a method first proposed by Vladimir Rokhlin of Yale University, but Chew's group helped pioneer the method, improvised it, and became the first ever to successfully use it for c omplex 3-D electromagnetic scatterer problems. They solved the scatterer equation for an aircraft codenamed VFY218, computed at 2 GHz frequency, the first success at this frequency.

The resources offered by NCSA were critical to the landmark success, Chew said.

"Without the access to the SGI Origin 2000 and parallel computer systems, this would have been impossible. NCSA allows us to solve these kinds of problems," he said.

Computational electromagnetics is becoming increasingly important in modern technology. It offers less expensive computer-aided design tools as replacements for many expensive experiments. In addition to calculating radar cross-sections, computational ele ctromagnetics impacts the design of computer chips, antenna and telecommunication systems, bioelectromagnetics, remote sensing, oil and mineral exploration, and micro electromechanical sensors.

The Center for Computational Electromagnetics collaborated with DEMACO in Champaign, IL, in a release of the computer software, called FISC (Fast Illinois Solver Code), last January. DEMACO has added a graphical user interface, and various geometry handling capability to the algori thm, making the code more user friendly. Even users with little or no electromagnetics background can use it to calculate the radar cross section of complex targets like aircraft.

Other professors working toward improving FISC are Professors Jianming Jin and Eric Michielssen. The Center for Computational Electromagnetics is currently supported by the Air Force Office of Scientific Research via the MURI program. More information abo ut the Center is available at http://www.ccem.uiuc.edu.

NCSA, a unit of the University of Illinois at Urbana-Champaign, is dedicated to advancing leading-edge technologies in information and high-performance computing and communications in academia and industry. Major funding to support its mission is received from NSF, the Defense Advanced Research Projects Agency, the DOD Modernization Program, NASA, other governmental agencies, NCSA's corporate partners, the state of Illinois, and the University of Illinois. NCSA recently received an NSF Partnership for Advanced Computational Infrastructure (PACI) award for it s proposal, the "National Computational Science Alliance", which becomes effective October 1, 1997.

CONTACT - John Melchi
NCSA public information officer
(217) 244-3049
jmelchi@ncsa.uiuc.edu
http://www.ncsa.uiuc.edu/General/PIO/