IT^2---Proposed Multidisciplinary Initiative in Information TechnologyMay 22, 1999
SIAM President Gilbert Strang devotes his column in this issue to two of the many large areas central to SIAM: computer science and computational science and engineering. In that context he refers, as he did in a previous column, to the IT^2 initiative.
What is IT^2, and how does it affect the SIAM community?
IT^2---Information Technology for the Twenty-first Century---is a proposed multiagency federal initiative, now making its way through Congress, created to support long-term, high-risk research in information technology. In addition to a change in emphasis for computer science research, IT^2, if implemented, would produce a major shift in the balance of funding among the disciplines supported by the National Science Foundation, which has been designated the lead agency.
The Clinton administration has requested $366 million for the initiative in its first year, 2000. (See sidebar below for details.) NSF's entire share ($146 million) would go to the Computer and Information Science and Engineering (CISE) Directorate, a 41.5% increase over the directorate's 1999 budget. Ruzena Bajcsy, a computer scientist from the University of Pennsylvania who works in robotics and, since January 1999, has been the head of CISE, has also been named head of the coordinating panel established to oversee IT^2.
The Department of Energy's response to the IT^2 challenge is the Strategic Simulation Initiative, according to Martha Krebs, director of the Office of Science at DOE. SSI seeks to revolutionize scientific research through the application of teraflop computation, which will require significant cross-cutting research in computer science and applied mathematics. SSI targets specific application areas, including global climate modeling and combustion simulation.
IT^2 is a direct response to the recommendations made by the President's Information Technology Advisory Committee (PITAC) in its interim report.* The formation of PITAC was itself recommended by the federal High Performance Computing and Communications program, which can be seen as a predecessor of IT*2. People working on the "grand challenge" applications of HPCC encountered obstacles, says PITAC member Joe Thompson of Mississippi State University. "IT^2 is a way to address the obstacles, to develop the enabling technologies."
Thompson points out that the "obstacles" he sees IT^2 addressing are cross-cutting---they don't arise in just a few applications, and surmounting them will require the efforts of multidisciplinary teams. The PITAC report, recognizing the multidisciplinary nature of much of the research required to meet the goals of IT^2, recommends an emphasis on broad projects of relatively long duration, with a renewed emphasis on research carried out by teams. The report also recommends the funding of collaborations with applications (with the information technology aspects of the research remaining the primary goal) and the creation of "Enabling Technology Centers" and centers for "Expeditions into the 21st Century." †
IT^2 is getting a lot of attention from the research community, for several reasons. The applied and computational mathematics community, Thompson points out, will clearly have a role to play in developing algorithms for the new technologies. Philippe Tondeur, who will become director of NSF's Division of Mathematical Sciences in July (see article), sees in IT^2 a wealth of mathematical ideas---geometric visualization, the study of algorithms, complexity theory, and data mining, among several others.
Nonetheless, concerns have been voiced. In a 1998 article, written for the Computing Research Association and published in slightly modified form in SIAM News (June 1998), SIAM past president John Guckenheimer anticipated some of those concerns. "Information technologies," he wrote, "are seen as a broader arena in which computer science is central. . . . It seems evident that critical research areas for computing are being neglected or de-emphasized." The numerical aspects of computing have no clear home in academic departments, he pointed out. Work on problem-solving environments and high-level languages is needed, but "we also need to maintain our investment in continuing research on [numerical] algorithms themselves."
Early in April, Ruzena Bajcsy met with a small group from SIAM, and on April 19 she presented a brief overview of IT^2 to the Joint Policy Board for Mathematics. Acknowledging that the channeling of all of NSF's funds through CISE "is causing some hiccups," she identified some needed contributions from the mathematical sciences. "We need your help in modeling large distributed systems," she said; "my community is worried-we don't understand network dynamics."
"You cannot build models and develop theory without understanding the algorithms," she said. "This [IT^2] has to be a partnership." "Where," asked SIAM JPBM representative Thomas Manteuffel, "will the people with the interdisciplinary background needed to work in such partnerships be trained?" "You tell me," Bajcsy replied, acknowledging the lack of easy solutions to the problem.
The job to be done under IT^2 is broad, Bajcsy said in conclusion. The hard problems include distributedness, asynchronicity, connectivity, and dynamic data structures. To succeed in solving these hard problems, "we need every neuron out there, and all neurons are not sitting in computer science departments."
*Established in 1997 to "assess the extent to which Federal support of fundamental research in computing is sufficient to maintain the Nation's critical leadership in this field," PITAC was chaired by Ken Kennedy of Rice University and Bill Joy of Sun Microsystems.
†Enabling Technology Centers are defined as "centers of excellence in computer science and engineering research applied to particular applications of information and communications technology." For the Expeditions---"centers, perhaps virtual, that bring together scientists, engineers, and computer scientists from academia, government, and industry to 'live in the technological future'"---the report cites a few existing examples: Xerox PARC and the MIT Media Lab.
The $366 million requested for IT^2 by the Clinton administration would be allocated in the first year as follows:
Long-term information technology research, $228 million
Advanced computing, $123 million
Social and economic aspects, $15 million
The $366 million would be distributed among the participating agencies as follows:
NSF, $146 million
Department of Defense, $100 million
Department of Energy, $70 million
NASA, $38 million
National Institutes of Health, $6 million
National Oceanic and Atmospheric Administration, $6 million
Important research topics identified in March by PITAC co-chair Ken Kennedy in testimony before the House science subcommittee:
Software (efficient methods for creating and maintaining high-quality software of all types, and for ensuring the reliability of existing complex software systems)
Scalable information infrastructure
High-end computing and communications
Elaborating on the high-end computing component, Kennedy described the need for "continued invention and innovation in the development of fast, powerful computing systems and the accompanying communication systems needed to implement critical science, engineering, and business applications ranging from aircraft design to weather and climate modeling. This includes research in innovative computing technologies, architectures and software for improving the performance of systems."
Kennedy likened the kinds of projects envisioned for IT^2 to those supported by DARPA through the 1980s.