CSE 2009: Graduate Education in CSE---Structure for the Zoo?
May 18, 2009
Participants in the well-attended Education Panel Discussion at the SIAM Conference on CSE, from left: Chris Johnson, Max Gunzburger, Michael Sch�fer, Ulrich R�de, Hans-Joachim Bungartz, Misha Kilmer, Lennart Edsberg, and Marek Behr.
Hans-Joachim Bungartz and Donald Estep
Many of us teach graduate students in computational science and engineering as part of our academic responsibilities, and we do so with considerable satisfaction. Most of us feel quite sure about the essential relevant content of the graduate curriculum, but at the same time are trapped within existing, and sometimes hindering, university structures.
Many institutions have created programs in CSE, which go by a number of names and which together offer quite a wide spectrum of solutions to this problem. For these reasons, the organizers of this year's SIAM conference on CSE included several sessions on graduate education in computational science and engineering: a two-session minisymposium, a panel discussion, and a student paper competition and prize.
In this short article, we discuss the minisymposium and the panel discussion.
The speakers in the first part of the mini-symposium---Donald Estep (Colorado State University), Celeste Sagui (North Carolina State University), Lennart Edsberg (KTH Stockholm), and Michael Sch�fer (TU Darmstadt)---provided an overview of existing programs and activities, worldwide and with a special focus on the U.S., Europe, or Germany. The second part---with presentations by Chris Johnson (University of Utah), David Keyes (Columbia University), Marek Behr (RWTH Aachen), and Max Gunzburger (Florida State University)---was devoted to issues important to programs everywhere, such as the handling of students from heterogeneous backgrounds, ingredients of CSE programs, specific tailored modules, beyond "a bit from here and a bit from there," and, last but not least, the possible role of SIAM as a promoter of CSE education.
After a short break, speakers and audience met again for the Education Panel Discussion. Perhaps a general remark to begin: A rule requiring at least a few panel discussions at all conferences seems to be in effect, and the SIAM CSE conferences ap-pear to follow it. Some of these events suffer from a rather obvious lack of interest; in many cases, the organizers' goal is simply to have more people in the audience than on the panel. With this in mind, we were a bit concerned, but the members of our panel---six of the minisymposium speakers or organizers, plus Ulrich R�de (Universit�t Erlangen-N�rnberg) and Misha Kilmer (Tufts University)---were far outnumbered by the audience. Donald Estep acted as moderator and guided panel and audience through an agenda of three main questions: (i) What are the characteristics and main ingredients of graduate programs in CSE? (ii) How can SIAM (better) support graduate education in CSE? (iii) What trends, needs, and threats lie in store for CSE programs and graduates?
Because of the broad spectrum of topics covered and opinions articulated, it is not easy to summarize this multi-faceted "educational afternoon" of the conference. Nevertheless, we would like to draw readers' attention to some of the general findings of the presentations and lively discussions:
- Data and related issues have gained in importance in relation to "mere" computations; this trend in CSE research will probably increase and must be reflected in educational programs.
- The adjective "interdisciplinary" has made its way into the mandatory vocabulary of most scientific areas, especially CSE. The still ubiquitous and powerful disciplinary "silos," however, frequently turn even the most eloquent statements into little more than lip service.
- Despite the appearance of a couple of reports and overviews of CSE education in the U.S. and elsewhere, a current version that would be continuously updated is lacking. This could define a possible role for SIAM.
- The bestowing of new names is frequently driven politically and thus probably inevitable, but to some extent this is not helpful for establishing our field---no one at the graduate education sessions, for example, could really articulate the distinction between CSE and the more recent "simulation-based science and engineering."
- Although courses are still and will probably remain a core component of graduate programs in CSE, there is a tendency toward more research-, project-, and team-oriented modules.
- CSE programs need both a top-down (boards, deans) and a bottom-up (experts) push; otherwise, the risk of failure is high.
- Finally, dedicated new schools or departments may be the path fraught with the greatest difficulty at the beginning, but payoffs come---at the latest---with the selection of new faculty. This was emphasized by David Keyes, who as department head of applied mathematics and computational science at King Abdullah University of Science and Technology (KAUST) in Saudi Arabia is part of
one of the world's most exciting current academic developments.
One issue that SIAM could or should take care of has already been identified: keeping track of worldwide developments in CSE education. Another closely related idea emerged from the discussions as well: What about guidelines for implementing programs in Computational X? In many countries, professional organizations or associations of university departments are doing this. For accreditation, for example, it is frequently important that the design of programs follow established recommendations. What is quite natural for physics, computer science, or electrical engineering, however, does not seem to apply to CSE so far. Although there was a broad consensus that differences in programs---arising from local boundary conditions, topical ancestors, or strongholds---should not be overly constrained, it is clear that some guidance would be helpful. And what group would be more ideally suited than SIAM to tackle this endeavor?
Together, the sessions in Miami made it obvious that educational issues of all kinds are a hot topic in CSE. Many problems were identified, but the story has an encouraging side too: The excellent quality of the contributions to the student paper competition clearly showed that graduate education in CSE is on a positive path.
Hans-Joachim Bungartz is a professor of scientific computing in the Department of Informatics at TU M�nchen. Donald Estep is a professor of mathematics and statistics and the director of the Center for Interdisciplinary Mathematics and Statistics at Colorado State University.
