To incorporate these views and objectives into a course, I have developed a series of projects which will be published by John Wiley & Sons as a traditional book, A Project Approach to Computation Physical Science, in the summer of 1996. Simultaneous with the development of the book materials has been the conversion of the projects into interactive tutorials for the World Wide Web. The tutorial development is supported by UCES (which has a slant towards undergraduate education)[1] and the Northwest Alliance for Computational Science and Engineering, NACSE[2] (which has a slant towards making high performance computing more accessible to scientists and engineers). Samples of both projects can be found at http://www webman@nacphy.physics.orst.edu/ .
As much as is possible, each project stresses the journey from physical problem to computational solution. To stimulate the students' curiosity and broaden the students' scope, we expose them to a large number of problems. To move at the required rapid pace, we provide elementary codes for the students to adapt for many of the projects, and discuss background material in lecture. We try to give the students the time and setting to understand--at their own rate--each project's virtues, areas of applicability, limits, and potential for visualization.
After discussions with an instructor, each student writes up the project as an ``executive summary'' focusing on purpose, theory, algorithm, code, and visualization. Recently I have encouraged the students to submit their reports as web document. I find that the hypertext format is excellent for our combination of text, code, graphs, and tables (not surprising since the world wide web was started at CERN to permit international collaboration in particle physics experiments). To further encourage a professional approach, I grade the contents and effectiveness of the report as if the student were presenting it to their boss at a weekly group meeting.
The topics covered are given in Table 
. They cover some
of the mathematics, computer science, and applications which I believe
computational scientists should know; in a traditional university
education these topics would be studied in a number of separate
courses. There are probably applications here which are unfamiliar to
computer scientists, and computer science and numerical methods which
are unfamiliar to many physical scientists. Yet when the student
works through all these subjects in a single course they see the
effectiveness of the algorithms, the power of the hardware and
software, and a glimpse into a simulated natural world they can
control.
The first quarter of the course concentrates on the basic mathematical, numerical, and conceptual elements needed for using computers as virtual scientific laboratories. After learning about Unix systems,[3, 4] we study the basics of computing: algorithms, precision, efficiency, and verification, and then move on to some numerical analysis and associated approximation and round-off errors. The second quarter focuses on realistic physical problems which apply and extend the preceding techniques. (There are more applications than can be studied in one quarter, and so some customizing is made to each student's interests.) An important aspect of the course is the use of advanced library routines, multiple-subroutine programs, interactions with larger research codes, and the use of supercomputers or parallel workstation clusters. This is often the only place students experience these common aspects of computational science.
Table: Topics in Computational Physics Course