CS and CS&EMay 22, 1999
From the SIAM President
I want to write this month about SIAM's work in two different areas with very similar names---Computer Science and Computational Science (and Engineering). I won't try to define them, because I couldn't do it. They don't represent all of SIAM by any means. We study dynamical systems and optimization and a tremendous range of applied analysis---plus models and algorithms to get solutions. But CS and CS&E are highly significant elements in the amazing (and I think wonderful) mixture that this society represents.
These two areas were highlighted for me in a recent day at the National Science Foundation. Four of us went to visit two directorates: Computer and Information Science and Engineering, and Mathematical and Physical Sciences. Going in, SIAM was already planning a special meeting in CS&E. Coming out, we seem likely to have a focused workshop in CS too.
Attentive readers of this column will know about the SIAM conference on computational science and engineering. This exceptional event, coming in September 2000 in Washington, will spotlight problems where computations truly support science. I almost wrote "big" computations and "big" science---problems like computational chemistry can be enormous. But the fundamental appeal for SIAM members is the challenge of producing the mathematical ideas and algorithms that those problems require. Often the crucial idea is very specific and seems pretty small. It is the consequences that are large.
Before leaving the topic of CS&E (to return in 2000), I want to emphasize that this conference is not a Washington initiative. The meeting is a SIAM initiative! The educational part, to develop applied mathematicians who connect directly to science and engineering, is now a widespread effort. Austria was an early leader (in Linz). Between the writing of this column and its publication, ETH Zurich will hold a workshop, planned by Rolf Jeltsch, on its new PhD program in CS&E. Similar plans are taking form in Scandinavia---and more places than I can name. An open discussion at ICIAM in July will report on many of them.
In computer science, SIAM's discussion at NSF was with Ruzena Bajscy. She and I became friends when she supervised Stéphane Mallat's remarkable thesis on wavelets at Penn. Now she is coordinating the work of six agencies on the Information Technology initiative, since NSF was designated to take the lead. We emphasized SIAM's important role in computer science, and also in scientific computing.
The conversation ranged over many topics. The one to write about here is Ruzena's suggestion of a workshop on basic directions in computer science---what problems should graduate students be attacking? SIAM has a lot of thinking to do, to make such a workshop successful. Your ideas are welcome. To give an example that I stumbled into, there are fascinating problems about file sizes on the Internet. The size distribution looks like a power law, with heavy tail and some self-similarity, definitely not Gaussian! A model that improves the packet-dropping protocol would be extremely valuable, as optical switching changes the rules.
May I end with another open question for SIAM? What role will we play in imaging and visualization? The work of many members-nobody knows how many-touches these areas. David Wilson from the University of Florida recently offered to organize a group in Imaging Science. It would be informal at first. In this subject and others, we are working next to giants like IEEE and ACM. But our emphasis is different. I am convinced that SIAM (and applied mathematics) can make a crucial contribution.