Details of Grant 

EPSRC Reference:	EP/C010027/1
Title:	Self-adapting software for Grid-based numerical simulation
Principal Investigator:	Jimack, Professor PK
Other Investigators:
Researcher Co-Investigators:
Project Partners:
Department:	Sch of Computing
Organisation:	University of Leeds
Scheme:	Standard Research (Pre-FEC)
Starts:	01 December 2005	Ends:	31 May 2009	Value (£):	187,450
EPSRC Research Topic Classifications:	Parallel Computing Software Engineering
EPSRC Industrial Sector Classifications:	No relevance to Underpinning Sectors
Related Grants:
Panel History:
Summary on Grant Application Form
Many important topics in science and engineering rely on computational simulation to improve our understanding and/or allow predictions to be made that would otherwise be unobtainable. Examples include the use of large parallel computers to undertake: (i) weather forcasting to predict short-term weather behaviour; (ii) climate modelling to understand and predict the long-term variations in the global climate; (iii) flow simulation to predict the aerodynamic characteristics of aircraft at the design stage without the need to build prototypes. There are many more such applications and they typically have a number of common features. For example, they generally involve teams of scientists or engineers who are distributed across different sites, and they frequently wish to use the most powerful computers that are available to them. The development of Computational Grids over recent years is aimed (at least in part) at allowing scientists and engineers, such as those described above, to work together more productively and more effectively by making computing and data storage facilities owned by different organisations (or different parts of an organisation) available in a secure, easy-to-use and accountable manner.In the past, when scientific teams have developed their simulation codes they have often done so with a particular target computer in mind (the one to which they have access) and so have selected their parallel solution algorithm, and tuned the various algorithmic parameters, specifically for this machine. With the advent of Computational Grids this approach is no longer viable since a given code will run on a variety of different architectures or, ideally, could be split between a number of different parallel computers (with different architectures). This is where the project proposed here fits in. The goal of this work is to move towards the situation whereby numerical simulation codes can be written so as to self-adapt to the architecture on which they find themselves running. This means selecting the most appropriate parallelisation strategy and automatically tuning the parameters associated with it so as to get the best possible computational performance. Furthermore if the behaviour of the computational resources or the network change during the simulation, or additional resources become available, then the software should be able to adapt dynamically to respond to these changes in its environment.The research that will be undertaken within this specific project cannot cover all posible numerical simulation scenarios so a number of important and typical examples will be selected as the basis for our initial work. The long-term goal would be to see self-adapting software being well understood and widely used for computationally intensive Grid applications within the next ten years.
Key Findings
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
Potential use in non-academic contexts
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
Impacts
Description This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk Summary Date Materialised
Sectors submitted by the Researcher
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
Project URL:
Further Information:
Organisation Website:	http://www.leeds.ac.uk