Academic Programs in Computational Science and Engineering

John R. Rice, Purdue University

T I n Ahmed Sameh’s and John Riganati’s article introducing this maga- zine in Computer (Oct. ’93, pp. 10-1 l), I presented a sidebar of informa- tion outlining 13 CSE programs. Now we take a closer look. While the programs’ names differ slightly, they are all t y n g to prepare their stu- dents to tackle the “grand computational challenges.”

Traditional computer science, physical science, and engineering pro- grams have not cross-trained their students beyond the college sopho- more level. T h e education that occurs beyond this level tends to be ad hoc, on the job, and self-taught. For instance, it is hard to find tradition- ally trained computer scientists who know enough about engineering and science to fully comprehend CSE applications; by far the most common educational path for such students avoids those disciplines entirely. In fact, CSE faculty often find that their PhD students have to spend an ex- tra year either learning about application areas or studylng topics weakly related to CSE; for example, abstract algebra, electrical power systems, or theoretical computer science.

CSE programs have risen out of a desire to remedy this situation. Al- though program specifics vary, the common thread is that they contain substantial computer science and engineering/science content. All the programs encompass more than one department, and most involve com- puter science. Ideally, students should learn most of the course material from two disciplines; since this is unreasonable, hard choices must be made as to what material to include. T h e following descriptions illus- trate the programs’ diversity as well as their common approach, combin- ing computer science, engineering, science, and applied mathematics in some way.

1070-9924/94/14.00 0 1994 IEEE 1 3

Clemson University

Clemson’s computational science program was developed in 1988 by R.M. Panoff (physics, now

at NCSA), D.E. Stevenson (com- ,

cLEhdsoN chemistry. Its philosophy is that computational sci-

ence is holistic: an umbrella for many disciplines rather than a discipline itself. The concept of computational science centers on applications, algorithms, and architectures.

Course work is aimed at the senior and post- graduate levels. One course, open to all science and engineering students, emphasizes interdisci- plinary group work through case studies. For ex- ample, last spring’s course had three groups among seven people: computer science and physics; physics and agricultural engineering; and computer science and textile chemistry. Clemson insists on this sort of arrangement to sensitize students to the problems of interdisci- plinary collaboration. The first course focuses on doing classical, easy-to-understand problems well. A second course is being developed to ad- dress advanced topics and advanced computer architectures.

Graduate students who want to concentrate on computational science must first be admitted to a specific department. This again emphasizes that computational science is only an umbrella. Computer science and the mathematical sci- ences offer a core of courses that prepare stu- dents in the relevant fundamentals. A doctoral candidate’s committee is chosen from among the faculty who are interested in computational sci- ence. T h e dissertation is expected to have a bearing on applications, algorithms, and archi- tectures. T h e program’s faculty run a weekly computational science seminar; recently, it fo- cused on development environments and soft- ware engineering issues.

The Clemson program is also active outside the university. Three faculty members partici- pate in the Undergraduate-Faculty Enhance- ment Program in Computational Science, which uses the facilities of the North Carolina Super- computing Center and the University of North

14

Carolina, Charlotte. T h e program targets schools in North and South Carolina, especially historically black and Native American institu- tions. Other projects in computational science include fostering the South Carolina Technical School’s ability to offer computational science support to state businesses.

For further information, contact D.E. Steven- son, Dept. of Computer Science, or D.D. Warner , Dept. of Mathematical Sciences, Clemson University, Clemson, S C 29634; steve@cs.clemson.edu, (803) 656-3444; o r warner@math.clemson.edu, (803) 656-3434.

George Mason University

Along with theory and experi- ment, computing is now part

of the triad that comprises methods of investigation; i t provides insight and leads to understanding that. in manv cases, theory and experiment cannot.

George Mason University established its Institute for

Computational Sciences and nformatics to address the role of

computation in science, mathematics, and engi- neering, and to sponsor multidisciplinary activi- ties that integrate various scientific fields. Rec- ognizing that computation is a central feature of its instructional and research program, the Insti- tute is in the process of establishing world-class computational facilities.

The doctoral and research program defines computational sciences as the systematic devel- opment and application of computing systems and computational solution techniques to mod- els that describe and simulate phenomena of sci- entific and engineering interest. Informatics is the systematic development and application of computing systems and computational solution techniques to experimentally, analytically, or bibliographically generated data to extract infor- mation of interest in science and engineering. The Institute hopes to produce new knowledge and understanding about, and approaches to, the research and educational possibilities to be found in nature’s complex systems.

The doctoral program emphasizes three ele- ments: common topics in computational sci- ences and informatics, computationally intensive courses in specific areas of interest, and doctoral research. Specialty areas include biology, chem- istry, computational fluid dynamics, earth sys- tems science, mathematics, physics, space

IEEE COMPUTATIONAL SCIENCE & ENGINEERING

sciences, and statistics. The program requires 72 graduate credit hours to earn a PhD in compu- tational sciences and informatics:

+ 12 hours of common computational sciences and informatics courses;

+ 12 hours from required courses in one area of interest;

+ 12 hours in electives from specialty courses in one area;

+ 12 hours from either additional computational topics, specialty research, individualized study based on professional experience and research, transfer credit, or other electives; and

+ 24 hours in dissertation research.

Students are encouraged to apply their knowl- edge to a broad range of scientific problems; they must acquire and use extensive computa- tional knowledge and techniques that are miss- ing from the more traditional science and math- ematics degree programs.

The Institute is a separate unit of the univer- sity and reports to the provost. It has its own faculty (currently six full-time and 10 part-time research faculty from other departments), plus associated faculty from 15 departments. The program is slated to expand to about 25 regular faculty and 200 to 250 doctoral students.

For further information, contact Menas Kafatos, Institute for Computational Sciences and Informatics, George Mason University, Fairfax, VA 22030-4444; phone (703) 993-3627; e-mail mkafatos8compton.gmu.edu.

Mississippi State University

In 1990, the National Science Foundation awarded Mississippi State University $1 2 million over five years to establish a research center for computational field simulation. The center was founded on two established research programs at Mississippi State: computational fluid dynam- ics and microelectronics design. The College of Engineering was then given permis- sion to award MS andPhD.de grees in both computational engineering and computer engineering.

T h e Engineering Research Center , now three years old, involves 29 faculty from nine de- partments, 25 full- time researchers, four postdoctoral researchers, 53 undergraduates, and 72 graduate students-17 of whom are in

computational engineering. All faculty, staff, and students have offices in the center, and a Sun workstation sits on each graduate student’s desk. There is also a lab/classroom of 18 Silicon Graphics workstations. The building is fully net- worked, and connected to Crays at the Naval Oceanographic Office, the Army Waterways Ex- periment Station, and the Mississippi Center for Supercomputing Research. The center has 17 industrial concerns and 14 federal labs or agen- cies as affiliates.

Students may enter Mississippi State Univer- sity’s graduate program in computational engi- neering after receiving an undergraduate degree in engineering, mathematics, computer science, or a physical science. The main thrust of this in- terdisciplinary program is the fusion of ideas from computer science and applied mathematics with a number of application areas. It also fo- cuses on modern computational techniques and numerical analysis.

Students are exposed to state-of-the-art nu- merical methods, high-performance computer architectures, use of software development tools for parallel and vector computers, and the appli- cation of these techniques to at least one scien- tific or engineering area. T o earn an MS or PhD degree, graduates must demonstrate expertise in an application area, such as computational fluid dynamics, computational electromagnetics, or computational heat transfer.

For further information, contact Jerry Rogers, Dept. of Electrical and Computer Engineering, Mississippi State University, Mississippi State, MS 39762; rogers@ee.msstate.edu, (601) 325- 3912.

North Carolina State University

National initiatives, local institu- - tional support, and faculty in- terest from several depart- . ments have propelled high-performance com- puting research and teaching at North Car- olina State University. T h e campus’s Center for Research in Scientific Com- putation (jointly formed by the 1 Computer Science and Math de- -

1

1

l-’ partments) acts as a focal point for degree- granting academic programs in scientific com- puting, including computational mathematics within the Mathematics Department and scien- tific computing within Computer Science. The two programs are very similar, and lead to

SPRING 1994 15

master’s and doctoral degrees in applied mathe- matics and in computer science, respectively. A third academic program, computational engi- neering and science, provides a structured, ex- panded, split minor in math and computer sci- ence and is available in all engineering and science departments at the master’s and doctoral levels. The university attributes the Center’s success to the cooperative efforts of its computer science and applied mathematics faculties de- spite being in different colleges.

Computer science is a vital component of the university’s scientific-computing research and teaching efforts. For example, of the 23 courses that support the computational engineering and science program, 18 are in computer science, with 8 of these cross-listed with mathematics. With proper advising, an informal scientific computation track is available within the com- puter science undergraduate program.

Several shared-memory and message-passing parallel computers are available on campus for researchers and graduate students, and a Cray Y- MP8/464 and a Kendall Square KSRZ-48 are at the North Carolina Supercomputing Center, which is located in Research Triangle Park and managed by the Microelectronics Center of North Carolina. A high-speed educational net- work connects NCSC to more than a dozen campuses.

For further information, contact Robert E. Funderlic, Dept. of Computer Science, North Carolina State University, Raleigh, N C 27695- 8206; ref@adm.csc.ncsu.edu, (919) 515-7775.

Purdue University

Purdue is establishing interdisciplinary master’s and doctoral degree programs in computational science and engineering. All 15 departments of

the School of Science and the Schools of Engineering are

involved, plus the depart- ments of Industrial and Physical Pharmacy, Medicinal Chemistry and Pharmacognosy, and Psychology.

Purdue’s CSE pro- gram gives students the

opportunity to study a spe- cific science or engineering dis-

cipline along with computer science in a multidisciplinary environment. The aim of the program is not to produce a student with parts of two degrees, but rather to train a student who has learned how to integrate computational

science with another scientific or engineering discipline. The expected course load and exams for students in this program are roughly the same as master’s or doctoral degrees in other disciplines at Purdue, with approximately one third of the course load and exam committees from the Computer Science Department and two thirds in the student’s home department (for students whose home department is Com- puter Science or Computer Engineering within Electrical Engineering, the reverse is true). Master’s degree students will be prepared to join and contribute to interdisciplinary re- search teams. Doctoral candidates are expected to become leaders in research and development at the forefront of their fields, applying ad- vanced computational techniques and theory to solve key problems. Additionally, the program’s planners hope it will foster interaction between faculty and students from the various depart- ments through colloquia and team research efforts.

Four courses form the core of the CSE part of the program: computational methods in linear algebra, computational methods in analy- sis, high-performance computing, and computa- tional science and engineering. The latter is a new graduate course surveying relevant material about computer systems, programming lan- guages, software engineering, and so on. CSE students also choose from among the existing, more specialized courses offered by various departments.

Parallel and vector computers in the Purdue University Computing Center, the Computer Sciences Department, and the School of Electri- cal Engineering support the program. In addi- tion, a CSE laboratory with multimedia work- stations will be established and partially dedicated to support the courses.

The program is governed by an advisory com- mittee made up of the heads of the participating departments, and operated by a graduate com- mittee of representatives from each department. This latter committee has subcommittees that carry out the various responsibilities, and the committee chair heads the program. The de- partments award the degrees and thus control the curriculum. Each graduating student’s tran- script contains both the field of study (the home department) and the area of specialization, which indicates that the student completed Pur- due’s interdisciplinary CSE program.

For further information, contact John R. Rice, Head, Dept. of Computer Sciences, Purdue University, W . Lafayette, IN 47907-1 398; jrr@cs.purdue.edu, (3 17) 494-6003.

16 IEEE COMPUTATIONAL SCIENCE & ENGINEERING

Rensselaer Polytechnic Institute Rice University

An interdisciplinary committee is organizing Rensselaer’s computational science and engineering program, which will begin admitting students in 1994. Commit tee members come from the follo ments: Computer Science; Electrical, Computer, and Sys- tems Engineering; Physics; Mechanical Engineering; Aeronautical Engineering and Mechanics; Chemistry; and Civil Engineering. T h e program is adminis- tered by Rensselaer’s interdisciplinary Scien- tific Computation Research Center. It will fo- cus initially on master’s and doctoral programs, but then introduce undergraduate concentrations and minors shortly thereafter.

CSE students will be associated with a home department within the science or engineering schools, according to their major interest. They will take a core of computer science and math- ematics courses, including numerical analysis and scientific computation. Beyond this, students will select a curriculum appropriate to their ma- jor field. Those in departments other than com- puter science or mathematics take about a quar- ter of their courses from the CSE core and about three quarters from among specialized science and engineering courses. Students having com- puter science or mathematics as home depart- ments follow a slightly different path, taking ap- proximately one quarter of their courses in an area of natural science or engineering.

Fifteen science and engineering faculty mem- bers belong to the Scientific Computation Re- search Center, which administers the program. Created in 1990 as a joint enterprise of Rensse- laer’s Schools of Science and Engineering, the research center maintains its mission to improve understanding of physical phenomena, provide new algorithms and solution techniques, and support computational experimentation. T h e group’s activities are varied, but the center’s central objective is to develop algorithms that reliably and automatically solve problems in- volving partial differential equations, parallel computational techniques and programming methodology, and optimal solution procedures for critical applications.

For further information, contact Joseph E. Flaherty, Amos Eaton Professor of Computer Science, Rensselaer Polytechnic Institute, Troy, NY 12 180; flaherje@lotus.cs.rpi.edu, (5 18) 276-6348.

As a consequence of the rapid increase in com- puting power over the past decade, modern sci- ence and engineering rely increasingly on com- putation as an aid to research, development, and design. Indeed, we can hardly imagine a large- scale engineering project that does not call upon some aspect of the mathematical and computa- tional sciences. However, using the newest and most powerful computers requires a knowledge of parallel and vector capabilities along with such things as visualization, networking, and programming environments. In addition, new algorithms and analytic techniques enhance the power of these computational tools. The obvi- ous relevance of these techniques to science and engineering led Rice University to establish a program in this area that can provide special- ized training in high-performance computing technology.

In conjunction with the Computer Science, Chemical Engineering, and Electrical Engineer- ing departments, the Mathematical Sciences De- partment has initiated a new degree program leading to master’s and doctoral degrees in CSE. The pro- gram focuses on modern computa- tional techniques and provides a re- source for training and expertise throughout the university.

The program is governed by T

a committee of faculty chosen by the dean of engineering, with ultimate oversight by the provost. This committee is re- sponsible for helping students to design appro- priate courses of study, setting examination re- quirements, and ensuring the integrity of the degree program. The committee is not a new department, but rather a mechanism for initiat- ing the interdisciplinary research required to ad- vance computational science.

The master’s degree is intended to produce experts in scientific computing who can work as part of interdisciplinary research teams. Graduates will be trained in state-of-the-art numerical methods, high-performance com- puter architectures, use of software develop- ment tools for parallel and vector computers, and the application of these techniques to a t least one scientific or engineering area. The curriculum consists of varied topics from mathematical sciences, computer science, and a selected application area. Requirements include successful completion of 30 semester hours or more of advanced courses (but no thesis).

SPRING 1994 17

Students design their course of study with the committee’s advice and approval.

It has been possible to construct this program from existing courses, with one exception. A new, one-semester course called Introduction to Computational Science now serves as the pro- gram’s core course. It provides an introductory survey of the topics that make up the scientific- computing program. Its purpose is to help stu- dents appreciate the program’s scope so they will be prepared to select their later courses. It also serves the broader needs of the engineering school and the physical sciences by introducing their students to state-of-the-art technology in scientific computing.

Students advance to doctoral candidacy by successfully completing a program of approved course work along with satisfactory performance on preliminary and qualifying examinations. The program adheres to the foreign-language requirements of the student’s department. Un- der the direction of a CSE faculty member, each doctoral candidate must complete an original thesis that is acceptable to the Computational Science Committee.

For further information, contact Danny Sorensen or Richard Tapia, Dept. of Mathemat- ical Sciences, Rice University, Houston, T X 77251; sorensen@rice.edu, (713) 285-5193; or rat@rice.edu, (713) 527-4049.

Stanford University

Since 1988, Stanford has had an interdiscipli- nary program in place for granting degrees in

scientific computing and computa- tional mathematics. The pro-

gram trains students to use \ advanced computer archi-

tectures and tools in vari- ous science and engineer- ing disciplines; its main

m T # thrust is the fusion of ideas from applied mathe-

matics, numerical analysis, omputer science, and several ication areas. With the advent

of modern computer architectures, i t seemed appropriate to develop a discipline that emphasizes the solution of problems requiring powerful computers in application areas.

T h e Scientific Computing and Compu- tational Mathematics program resides in the School of Engineering, and students are admit- ted directly into the program independent of other departments. The faculty, who are part of other departments, comprise three levels of par-

\ i

18

ticipation. Five core faculty members administer the program, while the associate faculty advise students and teach courses associated with their disciplines. There are also affiliated faculty who have less involvement in the program but teach courses of some interest to the students. There are now more than 30 PhD students, with back- grounds in mathematics, computer science, en- gineering, and medicine.

In addition, there are working relationships with local research centers such as the Research Institute in Advanced Computer Systems, Lawrence Livermore National Laboratory, and IBM.

Further information is available from Gene Golub, Dept. of Computer Sciences, Stanford University, Stanford, CA 94305; golub@pa- tience.stanford.edu, (41 5) 723-3 125.

Syracuse University

For the past three years, a group of faculty a t Syracuse have been developing programs in which undergraduate and graduate students can combine the study of computer science with an

engineering or scientific area. This new interdisciplinary area of study, called computational sci- ence, involves faculty from com- puter science, computer engineer- ing, mathematics, physics, chemistry, mechanical engineer- ing, and neuroscience. The pro- grams are administered by the School of Computer and Informa- tion Science in the College of En- gineering and Computer Science, and draw on many existing courses in applied mathematics, computa-

tional techniques for science and engineering, and computer science, especially those relating to high-performance computing. New courses that directly show the interplay between these topics have also been developed.

The Computational Science faculty plans to design a full range of academic programs. Cur- rently, undergraduate students in any field can earn an interdisciplinary minor in computational science. It involves an eight-credit, two-semester sequence of lecture and lab called Introduction to Computational Science and Scientific Pro- gramming, plus a t least 10 upper-division credits of electives, which may include a senior project. An MS degree program has been proposed; in the meantime, students can study essentially the same program under the interdisciplinary mas- ter’s program. There are three core courses:

IEEE COMPUTATIONAL SCIENCE & ENGINEERING

Introduction to Computational Science, Design and Analysis of Algorithms, and Methods of Nu- merical Analysis I. Courses are also required in Applications of Computational Science; Parallel Programming, Algorithms, and Architectures; Methodology and Techniques; High-Perfor- mance Software; and Advanced Computer Sci- ence and Software Engineering. In addition, graduate students in any degree program may receive a “Certificate in Computational Sci- ence,” which certifies that they have completed certain courses in this area. A doctoral certificate requires an additional elective, that the disserta- tion be supervised by a member of the Compu- tational Science faculty, and that the dissertation make a contribution to the field of computa- tional science.

The introductory courses are organized by modules covering typical application areas that require computation for solutions in particle sys- tems, field simulations, statistical techniques, and optimization problems. Each module in- cludes an introduction to the typical engineering or scientific problem, the numerical methods suitable for solving the problem, the computa- tional algorithms that carry out the method, the issues of high-performance computing for those algorithms, and how the resulting computation can be used in specific problems.

Another new gradate course, Case Studies in Computational Science, concentrates in more depth on computational techniques for statistical treatment of large data sets, computational fluid dynamics, and statistical physics.

The program is supported by the computing resources of the Northeast Parallel Architectures Center, which include a CMS, an SP1 with high-performance switching, two DECmpps, an nCube-2, and an iPSC/860.

For more information, contact Geoffrey Fox or Nancy McCracken, School of Computer and Information Science, Syracuse University, Syra- cuse, NY 13244; cps@npac.syr.edu.

The University of California at Davis

Computing has emergcd as a third way of doing science, complementing the time-honored the- oretical and experimental approaches. T h e computational approach to science has made significant contribu- tions in several disciplines such as aerodynamics, meteorology, and nuclear engineering, where previously in- tractable prob- lems have been

SPRING 1994

solved. Great promise and future growth lie in other disciplines such as molecular biology, ma- terials science, chemistry, and physics. The pos- sibilities opened up by the availability of high- speed computing have many similarities across scientific disciplines, so that a science of compu- tation does exist.

Questions of science, computational tech- niques, computer science, and mathematics are inseparable in addressing the large issues in computational science. A practitioner of compu- tational science must have skills in each area and be able to interact with people from all these ar- eas. With this philosophy in mind, UC Davis established a computational science program within the Applied Science and Chemistry de- partments.

The program is designed for graduate stu- dents who are interested in applying computers to the physical, chemical, mathematical, and en- gineering sciences. The program involves course work from the traditional areas of physics, chemistry, computational mathematics, and computer science, as well as in the area of the student’s specialization. Doctoral candidates in participating departments declare a designated emphasis in computational science, and take a special set of core courses in their home depart- ments and a set of core courses in computational science. For example, in the Department of Ap- plied Science, the core courses are applied math- ematics, computational mathematics, and a com- putational science course designed especially for physical scientists and covering such topics as computer architecture, parallel computers, algo- rithms, and numerical methods. After passing written examinations, students continue their graduate studies by taking electives from a vari- ety of courses in their department. Students are awarded a PhD in applied science with emphasis in computational science.

For additional information, contact Gary Rodrigue, Dept. of Applied Science, University of California a t Davis, Davis, CA 95616; rodri~ie~lIl-crg.llnl.gov, ( 5 10) 422-9787.

The University of California at San Diego

The discipline of scientific computation involves the formulation, analysis, and application of computational algorithms for the numerical so- lution of problems arising in science and engi- neering. An important characteristic of the disci- pline is the involvement of high-performance computing in both the theoretical and practical aspects of research. Scientific calculations gener-

19

ally take two forms, which we group together under the generic title scientific computation:

Computation involves the formulation, analysis, and application of computational al- gorithms for the numerical solution of math- ematical models; for example, finite-element methods for computational fluid dynamics, algorithms for image processing, or combi- natorial optimization algorithms for circuit design. Simulation includes the mathematical mod- eling of simplified processes in order to rep- resent real-world processes and systems on a computer; for example, rush-hour traffic in a metropolitan area, the real-time operations of a large telephone network, or the flow of air around an aircraft.

Computation plays a role in simulation, but the distinction between simulation and

enormous number of problems that were once considered intractable can now be solved. An inevitable consequence of this success is that the role of scientific computation in re- search has changed. It is no longer just an ad- junct to theoretical investigation, but now the principal means by which research is performed. Along with the ability to solve increasingly com- plex problems has come the need to train people in the science of computation. Ad hoc comput- ing techniques have evolved into sophisticated problem-solving tools that require a greater un- derstanding of computer hardware and software than was necessary in the past.

T h e planned PhD program (yet to be ap- proved by the University of California) focuses on areas of scientific computation that have sig- nificant overlap with the physical and mathe- matical sciences and engineering. Students will have to meet all the requirements of a home de- partment, but take a t least 20 units of elective courses from a scientific computation core. For the first year or two, the home department would provide basic training within the student’s major discipline. For the next one to two years,

the student would pursue a secondary specializa- tion and participate in a scientific computation seminar. In the final years, the student would complete dissertation research on a topic in sci- entific computation.

Students would enter the program through ad- mission to a participating department (currently including Applied Mechanics and Enpeering Sci- ences, Biology, Chemistry, Computer Science and Engineering, Mathematics, and Physics), which would serve as the home department and specify their primary specialization. Students would apply for admission to the Program in Scientific Compu- tation during the spring quarter of their first year. A typical PhD plan of study might include

+ Applied mechanics and engineering sciences: computational fluid dynamics, finite-element methods in solid mechanics, numerical meth- ods in engineering science, advanced com- puter graphics for engineers and scientists, and special topics in computational fluid dy- namics

+ Computer science and engineering: parallel computation, parallel and distributed compu- tation, system support for parallel scientific computation, and parallel algorithms

+ Numerical analysis/statistics: mathematical methods in physics and engineering, numeri- cal mathematics, numerical optimization, nu- merical partial differential equations, scientific computation, and applied statistics

For further information, contact Philip E. Gill, Dept. of Mathematics, University of Cali- fornia at San Diego, 9500 Gilman Drive, La Jolla, CA 92093-01 12; (619) 534-4879, peg@ optimal.ucsd.edu.

The University of Michigan

The University of Michigan offers a joint-de- gree doctoral program in scientific computing. The program is based on the recognition that a

firm knowledpe of the scientific application is an essential in-

cientific computation. Students are expected to complete the normal doctoral requirements for their home depart- ment, typically one of e traditional engineer- science, or mathematics nes, and take additional

gredient for research in

20 IEEE COMPUTATIONAL SCIENCE & ENGINEERING

courses in areas such as numerical analysis, sci- entific computation, applications, or the study of algorithms for advanced computer architectures. This interdisciplinary program is intended for students who will use these subjects extensively in their doctoral studies. The title of the degree has “and scientific computing” appended to it; for instance, a student might earn a doctorate in aerospace engineering and scientific computing.

The Laboratory for Scientific Computation administers the doctoral program in scientific computing, in cooperation with each student’s home department. Prospective doctoral students can pursue many different research topics in sci- entific computation, including

+ computational fluid dynamics, + simulation of VLSI circuits, + algorithms for advanced computer archi-

+ scientific visualization, + computational particle transport, + high-performance materials, + computational solid mechanics, + molecular dynamics, + simulation of semiconductor devices, + computational chemistry, + simulation of AIDS transmission, and + computer-aided molecular design.

tectures,

For additional information, contact William R. Martin, Director, Laboratory for Scientific Computation, University of Michigan, Ann Arbor, MI 48109-2 104; lasc-info@um.cc. umich.edu, (3 13) 764-5534.

The University of Utah

At the University of Utah, the departments of Computer Science (in the College of Engineer- ing) and Mathematics (in the College of Science) have recently developed a joint computational engineering and science program, designed mostly for students from the colleges of Engi- neering, Mines, and Science. First, it represents a meeting place to encourage interdisciplinary education and research among those who study and develop computational techniques for sci- ence and engineering applications. It is also a mechanism for students to obtain broader and more comprehensive training in computational science. A primary goal of the program is to train students in the use of advanced computing hardware and modern computational, graphical, and mathematical techniques for the solution of science and engineering problems that are inac-

cessible without such integrated ex- pertise. T o acknowledge successful program completion, the university will issue a certificate in addition to the regular gradu- ate degree. At a later stage, the program may grow into an interdiscipli- nary and cross-college graduate - degree program.

T o obtain this graduate certificate, a student must complete courses in architectures and algo- rithms, numerical analysis and computation, ad- vanced numerical analysis and computation, sci- entific visualization, mathematical modeling, case studies, and a seminar, plus complete a project in an application area outside of mathe- matics and computer science (this will normally be satisfied by the student’s thesis). The Com- puter Science and Mathematics departments re- cently created new laboratories consisting of several high-end graphics workstations and servers joined by high-capacity network links. The new labs augment other computing facili- ties throughout the university as well as a t the Utah Supercomputing Institute.

For more information, contact Chris Johnson, Dept. of Computer Science, or Peter Alfeld, Dept. of Mathematics, University of Utah, Salt Lake City, UT, 84112; crj@cs.utah.edu, (801) 581-7705; or alfeld@math.utah.edu, (801) 581-6842. 4

Acknowledgment

I thank the contact people who provided the initial drafts of these program descriptions.

John R. Rice is W. Brooks Fortune Professor of Computer Sciences and heads the Computer Sciences Department at Purdue University. His research interests focus on methods for solving partial differen- tial equations and on high-level, powerful systems for scientific com- puting. He is now leading the effort to establish a graduate degree pro-

gram in CSE at Purdue involving as many as 20 depart- ments and 100 faculty.

Rice received his BS and MS from Oklahoma State University (then Oklahoma A&M) and his PhD from the California Institute of Technology, all in mathematics. He is this magazine’s area editor for problem-solving en- vironments, and a member of ACM, the IEEE Computer Society, IMACS, and SIAM. He is also past chair and a member of the Board of Directors of the Computing Re- search Association.

Rice can be reached a t the address listed under Purdue University.

Y

SPRING 1994 21

______