Research Interests

Past Projects and Interests

The research areas listed in this section are mainly things I am no longer actively working on, though this doesn't mean I am no longer interested in these topics.

Performance evaluation of high performance computers.
Publications.
Particle-in-cell algorithms on parallel computers.
Publications.
Numerical linear algebra libraries for parallel computers, in particular the ScaLAPACK project.
Publications.
In 1994-5 I did some work with Professor Brian Straughan on eigenvalue problems arising from hydrodynamic stability computations.
I was closely involved in the development of the MPI message-passing library.
Publications.

Current Projects and Interests

The research areas listed in this section are those that I am currently working on. The first four topics are related to distributed high performance computing, and the fourth to mathematical modelling.

Problem-solving environments
A Problem-Solving Environment (PSE) is a computing environment that provides all the resources needed to solve a certain class of problem. PSEs may be used for modelling and simulation, for controlling scientific instruments and sensors, for optimising the design of industrial processes and products, and for navigating and analysing large, complex data sets. There is an increasing trend for the resources accessed by a PSE to be distributed over a network, and PSEs often support interactivity, collaborative working, and immersive visualisation.
A PSE is intended to provide the end-user with an interface that hides the complexities of the underlying hardware and software environment. This allows the end-user to focus on the problem to be studied without needing to be concerned with the low-level details of the resources used. Thus, through a PSE, an end-user can address a problem in the language of his or her own discipline, rather than needing to know also about computer operating systems, parallel and distributed computing, numerical algorithms, and other areas that are not of immediate concern.
My research into PSEs is currently funded by the EPSRC and by Britsh Aerospace. My PSE research group web page describes this work, and contains links to other research in this area.
The European Research Conference on PSEs took place in Spain in June 1999.
The EuroTools Special Interest Group on PSEs is a focus for research into PSEs in Europe.
Publications.
Java for High Performance Computing
Since its inception the Java programming language has rapidly become increasingly popular. Foremost among its attractive features is the promise of platform independent programming - a Java code will run (with a few caveats) on any machine with a Java Virtual Machine (JVM) resident. Other attractive features stem from Java's object oriented programming model, such as modularity, maintainability, and the ability to reuse software components. Furthermore Java's automatic memory management, operating system abstractions, and C-like syntax make it easy to learn and use.
In the past few years, Java has progressed from being used mainly for graphical enhancements to Web pages and GUIs to being a mainstream programming language. However, it has not been widely adopted yet for high performance scientific computation. Among the reasons for this are Java's lack of support for numerical computing, and its generally poor performance compared with C and Fortran. Scientific computing generally requires high performance, and a programming language that cannot provide it will not be used regardless of how good it is otherwise.
I am interested in investigating and improving Java's usefulness for high performance scientific computing, particularly in distributed environments. To this end I am involved in the European JavaGrande Forum. I also helped organise the First UK Workshop on Java for High Performance Network Computing which was held at EuroPar98, September 2-3, 1998.
Publications.
Digital Libraries of Remote Sensing Data
This research project, which began in July 1999, is investigating the use of a mobile agent technology for providing distributed access to, and parallel processing of, a digital library of Synthetic Aperture Radar (SAR) data.
This work is being done in collaloration with Dr. Roy Williams of the California Institute of Technology and Professor Giovanni Aloisio of the University of Lecce, Italy, and is based on the SARA project.
Check out the article in the Agent Link newsletter on "A Multi-Agent System for Analysing Synthetic Aperture Radar Atlas (SARA) Data" by Omer F. Rana, Yanyang Yang, Christos Georgesoupolou, David W. Walker and Roy Williams.
Publications.
Mobile Agents for Distributed Computing
The use of mobile and static agents in distributed computing environments features in all three of the above ongoing projects, so although I don't have a separate agents project, this is another area in which I am interested.
I was involved in organising the Workshop on Agent-based High Performance Computing at the Autonomous Agents 99 conference, and the Workshop on Infrastructure for Multi-Agent Systems at the Autonomous Agents 2000 conference.
Publications.
Modelling Flow in Fractured Media
The aim of this work is to develop mathematical models of colloid-mediated contaminant transport in fractured media. This is important in the area of groundwater cpontamination since pollutants adsorbed onto the surface of colloids can be transported more rapidly by flows through rock fractures than through the rock matrix. The objective of the work is to develop a simple analytic or semi-analytic model for the flow in a single fracture, and to use this as a basis for the flow in a network of fractures.
Publications.

Future Projects and Interests

The topics listed in this section are areas that I am interested in, but have not yet had time to develop into funded projects, or to do very much work on, but which I would like to pursue research into in the future, or which I am doing just as a hobby.

Lattice Gas Algorithms
I am interested in the use of lattice gas algorithms (LGAs) for modelling fluid flows, and more generally in the use of cellular automata for doing computational science and engineering. So far I have written a code for running FHP LGA models that models flows with fairly simple geometries and boundary conditions.
Driven cavity
http://www.cs.cf.ac.uk/User/David.W.Walker/research.html
David W. Walker
Email