D. E. Stevenson Research Interests

This Page Under Major Reconstruction

Index

General Statement of Research Interests

Professor Stevenson is interested in the development of a computational science and engineering (CS&E) infrastructure in the K--12, university, and industrial settings. Clemson computer science and mathematical sciences have been active in CS&E from the very beginning. For early information see John Rice's "Academic Programs in Computational Science and Engineering" (IEEE Computational Science & Engineering, 1(1). Spring 1994). We have an outdated CS&E home page. Other organized involvement in CS&E education is through the Shodor Education Foundation

The scientific fundamentals of CS&E were first laid out in "Science, Computational Science, and Computer Science: At a crossroads" (Communications of the ACM. 37(12). 85-96.).

Computational Science and Engineering

Computational science and engineering studies the proper conduct of science and engineering on computers. There are three basic aspects based on the original ACM article.:

  1. Application.

  2. Algorithms.

  3. Architectures.

We have come a long way since our first CSE paper in 1991. In October, 2002, Shodor Education Foundation, Inc. received a $2.75M grant to promote computational science education nationwide through a series of workshops. The schedule and other pertinent information is on the National Computational Science Institute website.

This picture remains correct even if we extend CS&E beyond science and engineering. A new model is hinted at in my article "How Goes CS&E : The First IEEE Workshop Revisited" in IEEE Computational Science and Engineering, Summer, 1997. The proposed model is P3:

  1. Paradigms,

  2. Problems, and

  3. comPutable solutions.

Implementation of CSE in Universities.

Papers of interest in the area for curricular development are

Questions of Quality

Why should you care? After all, isn't this just academic babble?

There are ample indications that the quality of software just isn't what it should be. The most famous and accessible listing of the indictments against computer science quality are found in the work of Peter Neuman. This work leads to the work of people like Nancy Leveson. Many of their stories are not related to the unique problems of CSE. I have tried to pull some stories together in my Stories file (Postscript version). Any and all contributions gratefully accepted.

The question of quality is a deep one. Supposedly, software engineers worry about such things. However, the methods of the software engineer do not jibe well with the methods of the physicist, engineer, or mathematician. Therefore, new methods are needed. One take is given in my submitted article "A Critical Look at Design, Verification, and Validation of Large Scale Simulations " to appear in IEEE Computational Science and Engineering

I have some basic ideas about V&V that I wanted to try out on the Michelson-Morley experiment of the 1880s. The write up for the European Simulation Multiconference is now available. The slides are also available.

The ongoing saga of VV&A. There are, of course, real people doing VV&A everyday. The Defense Modeling and Simulation Office (DMSO) is one clearing house. But how to use all the stuff people are saying? Here's an example, based on the Michelson-Morley paper noted above

Building on the Michelson-Morley experience, we have developed a variant of some work done by Good, Polya, and Jaynes. This foundation is termed An Evidence-Based Approach to Fidelity. Note that fidelity is the term used in the defense establishment, whereas science uses the term validation.

Going back to the well, a much expanded version of the Michelson-Morley experiment is available as a pre-print.

One of the questions for V&V is whether or not we can use formal methods. The Verificaton and Validation of Complex Systems paper points out that verification and validation are formally related.

The Foundations '02 Workshop, Foundations for Verification and Validation in the 21st Century, was held at Johns Hopkins Applied Physics Laboratory on October 22--24, 2002. A review of the research problems discussed is in SCSC '03. The proceedings are available from the Defense Modeling and Simulation Office and mirrored at Clemson University. The plans for A Foundations '04 are in the works. Information on Foundations '04 are available.

The DEVS formalism of Bernard Zeigler, et al., has a natural formulation in category theory. The first vesion of this formulation is to appear at the SCSC '03

The Philosophy of Computational Science

A study of the foundations of computation is in order. There are two questions to look at. I take the view that the "classical" Turing-Kleene view is not a correct one and that we should demand constructive and not classical approaches only. This view is expounded in What is Computational Knowledge and How Do We Acquire It?

The philosophy only comes after one has investigate the basis of it all. But where do the basic tenets of computation come from? The begin at the beginning means reinterpreting Euclid. Could Euclid be constructive? See Principles of Constructive Euclidean Geometry

Can Philosophy Actually Be Useful?

The reason to study foundations is to come to understand how the "practitioner" (software engineers, programmers, and users) come to determine the correctness of computer systems. We're not talking testing or proving programs here: we're talking about actually doing something. Here's the text of my talk at Sandia National Labs title Validation and Verification: A CSE Perspective.

Two talks, a bit dated, recently given at the University of Stockholm, Stockholm, Sweden. The slides for those presentations are available:

The full presentationwas submitted but the special issue of the JSC was called off

Papers

 An evolving list of special interest journals.
Steve Stevenson
Last modified: Wed Jun 18 11:50:26 EDT 2003