Computational Science and Engineering at UC Davis

Final Report of the Committee on
Computational Science and Engineering





Committee Chair:
Joel E. Keizer, Institute of Theoretical Dynamics

Acting Committee Chair:
Bernd Hamann, Ctr. for Image Processing & Integrated Computing

Other Committee Members:
Harry A. Dwyer, Mechanical and Aeronautical Engineering
Daniel Gusfield, Computer Science
Alan M. Hastings, Environmental Science and Policy
Winston Ko, Physics
C. William McCurdy, Jr., Lawrence Berkeley Lab., Applied Science
David M. Rocke, Graduate School of Management
William P. Thurston, Mathematics

June 10, 1999

Preamble

This Report concerns the future of computational science and engineering (CS&E) at the University of California, Davis. CS&E is a rapidly developing area with particularly strong connections to the sciences, engineering, and mathematics. CS&E is concerned with the development of computational models as an alternative way to help understand complex physical and biological processes--or to model entirely abstract processes, encountered in mathematics and computer science. CS&E involves diverse areas such as Monte Carlo simulations of random processes; algorithm development for the analysis of massive and multidimensional data sets; numerical algorithms for the solution of differential equations characterizing a wide variety of physical phenomena; development of methods for the analysis of biomedical diagnostics; computational molecular biology; visualization and virtual reality rendering for the study of large and complicated three-dimensional structures; digital image analysis, compression, and transmission; and computational techniques in relation to the study of discrete mathematics, including cryptography and combinatorics. A significant portion of UC Davis faculty and researchers is actively involved in a variety of CS&E fields and the development of algorithms and methods for solving large-scale problems impacting the future of the research process and teaching in science and engineering. It should be the objective of this Initiative to create an environment at UC Davis that will enable world-class education and research in CS&E. The Committee is convinced that CS&E will play an important if not a dominating role for the future of the scientific discovery process and engineering design. We provide three examples:

• Computational Physics. National needs in the areas of numerical simulation, data exploration, and large-scale high-bandwidth communication have led to the creation of the so-called Accelerated Strategic Computing Initiative (ASCI) of the Department of Energy. Three major DoE laboratories--Lawrence Livermore National Laboratory (LLNL), Los Alamos National Laboratory (LANL), and Sandia National Laboratory (SNL)--are spearheading revolutionary computational methods to simulate complex physical processes and the behavior of materials under extreme conditions. CS&E is vital for the success of ASCI¹.
• Computational Neuroscience. With today's high-resolution imaging technology it is possible to produce brain scans that reveal thoughts, moods, and memories--as clearly as a traditional X-ray image reveals bone structure. We are now able to observe a person's brain registering an event or experiencing a painful memory. We can generate a static map of the human brain and can observe the electrical processes as they occur over time. Neuroscientists are beginning to develop models of the human brain that are direct results of this latest high-resolution imaging technology. The next step is the development of algorithms that simulate the observed brain activity patterns. Again, computational science and engineering methods will play a key role in advancing this area of research.
• Computational Biology. One of the most complicated processes currently being studied by scientists in a variety of disciplines, and primarily in molecular biology, is the so-called folding problem. The molecular structure of complex protein molecules is believed to greatly impact the eventual function of these proteins. It is crucial to understand the time-varying process of process folding itself, and computational techniques are of fundamental interest to (computational) molecular biologists to better understand the relationship between the structure of protein molecules, the folding process that leads to a final structure, and the function of the protein.

These examples do not represent the Committee's ``priority areas'' for CS&E, but they demonstrate that computation is becoming a key element in diverse fields. While most UC Davis disciplines that would greatly benefit from computation are not yet fully utilizing its power, we believe that the teaching and research components of this Initiative will help craft the kind of computation-oriented environment that will allow our university to remain competitive and become a national leader in CS&E in the next century. It is obvious that computational methods are a key component of the modern research process throughout the science and engineering fields. To be competitive as a research university in the ${21}^{\rm st}$ century, UC Davis will have to make significant investments in its computational infrastructure enabling efficient research on large-scale problems. The research aspect of this Initiative is to be viewed as at least as important as the teaching component. The teaching, development, study, and application of computational methods is already crucial for a large variety of disciplines at UC Davis. To ensure the competitiveness of UC Davis in science and engineering education and research UC Davis must substantially strengthen and focus its efforts concerning CS&E in the coming years. The main objective of the CS&E Initiative must be the creation of an educational and research environment that is both attractive and accessible to students², current faculty, and faculty to be hired. Linking this objective to the established and highly visible programs of UC Davis in the sciences and engineering should allow a maximum degree of leverage through potential joint appointments for new faculty hires. One must be careful not to confuse CS&E with the established discipline of computer science, with its focus on the organization, design, analysis, theory, programming, and application of digital computers and computing systems³. Nonetheless, the rapid growth of CS&E has been triggered by key hardware and software developments in computer science and engineering. It will be crucial for the success of the CS&E Initiative to foster a strong relationship with the Computer Science Department as well. Computer Science will play a significant role in the basic preparation of both graduate and undergraduate students minoring or specializing in CS&E. Moreover, the Computer Science Department already has a dedicated sub-group of faculty working on CS&E problems (e.g., visualization/computer graphics, computational biology, medical informatics, database/information systems, combinatorial optimization, cryptography, and other fields) and thus would be a natural home for several of the FTEs or joint CS&E appointments resulting from the Initiative.

To accomplish the desired major impact of CS&E on the campus, it will be crucial to have a strong interaction with all CS&E-impacted and -interested science and engineering units. CS&E is a field that must not be viewed in isolation from applications or as a field that could thrive by merely looking inward. CS&E must reach out to science and engineering in order to justify its existence, improve and strengthen ongoing computing-based efforts, and promote an increase in the utilization of CS&E across science and engineering in general. It is the view of the majority of the Committee members that UC Davis needs a CS&E facility and that it should couple CS&E activities and the tackled science and engineering applications very closely. Without this coupling there would be the danger that departments and units across campus might initiate their own independent CS&E-like activities, and this would surely deprive a potential CS&E unit of students, interactions, and funding. The development of CS&E on campus must not be done in isolation from the campus community at large. Only this will ensure the success of a centralized CS&E effort, an effort that would be accepted broadly, would minimize the risk of duplicating efforts, and would lead to the desired interdisciplinary faculty interactions in CS&E.

Another important aspect of the development of CS&E relates to the existing Graduate Group in Applied Mathematics (GGAM). This Graduate Group consists of approximately sixty faculty members and was formed to train applied mathematicians to carry out research in the physical sciences and engineering. GGAM aims at satisfying the need for applied mathematics education as it relates to science and engineering applications. GGAM offers MS and PhD degrees, and students may be supervised by any faculty member of the Group. The areas represented by this Group span a wide variety of fields, currently including population biology, atmospheric sciences, continuum mechanics, optimization and control, theoretical chemistry, computer and engineering sciences, mathematical physics, scientific visualization and geometric modeling, and mathematics. GGAM should play an important role in the development of CS&E. The reason for this is that the ``three pillars'' of CS&E are (i) science and engineering applications; (ii) methods of applied mathematics; and (iii) computer science techniques for practical algorithm implementation. The development of CS&E should build on these three components and ensure that more and improved interactions will result. Moreover, the Graduate Group in Statistics could play a similarly important role in this context. Both applied mathematics and statistics principles are of crucial importance for algorithm development to solve CS&E problems. There is no lack of funding opportunities in CS&E. Examples are large-scale programs such as the Accelerated Strategic Computing Initiative (ASCI), the Next Generation Internet Initiative, and the Information Technology for the ${21}^{\rm st}$ Century program. ASCI is the result of a nuclear test ban treaty prohibiting the U.S. from testing nuclear devices. ASCI is concerned with the development of large-scale CS&E technology allowing the computer simulation of the behavior of nuclear devices. ASCI and the other cited programs require a major involvement of academic institutions, which is a significant asset concerning the development of CS&E at UC Davis.