CSE Community Speaks Out in Policy ArenaJune 25, 2004
Talk of the Society
SIAM has grown to represent a rather broad community, although in truth a diversity of interests has characterized SIAM almost from the beginning. The interests of our members span a spectrum from theoretical applicable mathematics, topics like the analysis of PDEs or discrete mathematics, to more specific modeling of physical systems (or even nonphysical systems, as encountered in economics or the behavioral sciences). Of course, much (but not all) of the work our members do has a strong numerical or computational component.
The interest in the applications is the glue that binds applied mathematics together, whether that interest drives new areas of theoretical research or leads to codes that implement computational simulations. This interest cuts across the academia-industry lines, and encompasses many academic departments.
Still, from time to time a SIAM member comes forward with concerns that we emphasize one topic too much, or don't pay enough attention to a certain area of interest to some in the SIAM community. The fix for this, of course, is for such people to get involved and to push for the kinds of activities they would like to see.
Eventful Times for CSE
One topic of interest to a substantial part of the SIAM community---computational science and engineering---is also receiving a great deal of attention in Washington these days. CSE, which lies at the intersection of applied mathematics, computer science, and a specific application domain (think "computational x," where x is an area of science or engineering), embraces scientific computing as a component.
Several SIAM members put vast amounts of talent and energy into SIAM activities in CSE. The officers of SIAG/CSE, for example, are Lori Freitag Diachin, Jim Glimm, Omar Ghattas (who is also editor-in-chief of SIAM's book series in CSE), and Tammy Kolda (whose article on an upsurgence of interest in parallel processing appears in this issue of SIAM News). The editors-in-chief of SIAM journals important to the area include Tom Hou (MMS), Howard Elman (SISC), and Max Gunzburger (SINUM).
Some of the SIAM leaders in CSE have also taken active roles in a wider arena, interacting with a fairly broad range of agencies, from the National Science Foundation to the White House Office of Science and Technology Policy (OSTP), as well as with a variety of congressional committees.
The SIAM Committee on Science Policy has an abundance of such leaders, beginning with its current chair, SIAM past president Tom Manteuffel. At its spring meeting, the CSP heard presentations by key staffers from OSTP (which is the office of the president's science adviser). That meeting, in turn, led to a visit to OSTP, where two members of the CSP---David Keyes and SIAM president Mac Hyman---discussed research in applied mathematics and computational science before a larger set of OSTP staff.
Before looking at that very constructive meeting, a note about the efforts of another member of the CSP: Alan Laub writes that a version of the long-awaited "HECRTF Federal Plan" (the report of the 12-agency High-End Computing Revitalization Task Force, which he co-chaired) has been released to the public; we encourage readers to check it out (http://www.nitrd.gov/pubs/2004_hecrtf/20040510_hecrtf.pdf). We couldn't agree more with Laub's hope for the result of the task force's work: "It's going to be the start of what I hope is a true revitalization of high-end computing in the U.S.---along with all the math and CS that goes with it!"
At OSTP, David Keyes focused on the role played by algorithms, along with continued advances in hardware, in advancing computing capabilities. This is a presentation that he has sharpened and polished in the course of delivering it many times, before a variety of audiences. Participants in the March 2003 SIAM workshop on CSE, mathematics, and computer science (chaired by Jim Hack, Juan Meza, and Margaret Wright) and the more recent (April 2004) NSF/MPS Cyberscience Workshop heard it at various stages of its evolution.
Keyes acknowledges the crucial role of hardware---clearly, we need the best hardware capabilities available if we are to continue the dramatic increases in capability that computational modeling and simulation provide. But the point he emphasizes is that advances in algorithms deliver equally impressive advances-and often make it possible to solve problems that would not be tractable otherwise.
Citing examples from the SCaLeS (Science Case for Large-scale Simulation) report (developed for the Department of Energy; http://www.pnl.gov/scales/), he tells the story of increases in effective sustained speeds in classes of codes (like magnetohydrodynamic, or MHD, simulations) by nearly five orders of magnitude over two decades (see accompanying slide from his presentation). Advances in hardware contributed to some of this progress (about two orders of magnitude), but equal credit goes to improved algorithms-including partial-ly implicit schemes (filtering the fastest waves), semi-implicit schemes (treating all waves implicitly), use of higher-order elements, and improved linear solvers.
Keyes compresses all this in a sound bite: "The Gordon Bell Prize outpaces Moore's Law." It's an effective way to make the point that increases in computing capability come from faster hardware and improved algorithms---we need both---and from advances in modeling. At this point in the discussions at OSTP, Mac Hyman stepped in with a convincing case that improved modeling opens up entirely new capabilities, leading to understanding, analysis, and compu-tational simulation of new phenomena.
The SIAM community has long been a strong advocate for applied mathematics and computational science, and we are pleased at the growing recognition among policy makers of the value of research in these areas.
Communicating with Congress
Culminating the growing awareness in the U.S. of the need for high-end computing capabilities, efforts are now under way in the House of Representatives to reauthorize HPCC legislation---the High-Performance Computing Revitalization Act of 2004 (HR 4218). Many readers will remember the original form of this legislation, which led in 1992 to the coordination of federal investments in high-performance computing through a road map (the so-called "Blue Book") for research investments.
The proposed new legislation calls for providing "sustained access" to world-class computing systems-a vital component-and also encourages federal agencies to include support for other components critical to high-end computing projects: applied mathematics and computational science (among others), as well as software development. This is important recognition that modeling and algorithms are essential partners in the development of new capabilities.
A discussion of emerging leaders and spokespersons from the community wouldn't be complete without mention of recent outside recognition of some prominent members of our community. This spring, the National Academy of Sciences announced the names of those elected to membership this year. Among them are Phillip Colella of the Applied Numerical Algorithms Group at Lawrence Berkeley National Laboratory; an active participant in SIAM activities, he works in the area of scientific computing and computational fluid dynamics. Also elected were Stephen H. Davis of Northwestern University, who works in the area of interfacial dynamics and stability, and MIT computer scientists Shafrira Goldwasser and Ronald Rivest, both known for their contributions to theoretical computer science and cryptography.
Other new members from the mathematical sciences community include Benedict Gross of Harvard, Nicholas Katz of Princeton, and Charles Newman of the Courant Institute.
On a completely different note, many organizations, including SIAM, have expressed concerns about current visa policies and their potential impact on science in the U.S. More extensive background checks, instituted in the wake of the 9/11 attacks, have led in certain cases to extensive delays for students entering the U.S. and for non-U.S. citizens traveling to scientific conferences in the U.S.
The U.S. National Academies have called for a special visa "for foreign scientists and engineers who are well known to the research community, who have proven track records as participants in international research activities, and who need to make multiple visits to the United States." According to the NAS and NAE, "Such scientists and engineers should be eligible for long-term, multiple-entry visas; and they should be able to revalidate those visas from within the United States." Readers encountering difficulties are encouraged to consult the NAS/NAE Web site on the subject (http://www7.nationalacademies.org/visas/).
SIAM is working to inform policy makers of the concerns of the community and of the threat posed by the new policies to scientific research. We have endorsed a policy of open communication in basic research.