Specialist Themes

MeSC focuses on "Modelling and Analysis of Large Complex Systems" as its area of excellence.

Large complex systems arise in all areas of science and engineering, be it in pure science (nuclear physics, gravitational modelling, molecular modelling, or algebra), applied science and engineering (modelling of road and computer network traffic, climate, movement of earth structures, automated designs, factory processes) and the life sciences (the genome project, social interactions and crowd behaviour). The need for their computational modelling and analysis may arise through curiosity-driven fundamental research, such as in physics, mathematics or genetics, or the desire to predict the behaviour of systems being designed (road systems, networks, electronic devices, machinery, factory processes) or perhaps natural phenomena (earthquakes and weather patterns). The sources of complexity vary, ranging from very large and possibly geographically distributed data sets, through large numbers of components and complex interaction patterns, to the need to model complex evolution over time or in 3D space. Inevitably, the increase in complexity of systems being studied is not keeping up with the advances in computing, meaning that in order to obtain timely results, or even any results at all, it is necessary to harness the power of many computers at once. The Grid provides an unrivalled opportunity to make advances in the modeling and analysis of large-scale systems. It readily adapts to tasks involving large distributed data sets, but the performance is less predictable in cases where the computational processes that result are highly concurrent and require synchronization. The research challenge to make computation using the Grid effective is subject to multi-billion dollar effort world-wide and in the UK, with specific challenges emerging for computer science as recently identified by UK computer science experts.

There are four specialist themes of research in the Centre:

Simulation of evolving systems of interacting elements

This theme focuses on problems where the computational task is to simulate the dynamics of systems comprised of large numbers of interacting computational elements to observe their emergent behaviour. The theme encompasses a wide spectrum of simulation approaches, from discrete element method to cellular automata, neural networks and agent-based discrete event simulations.

Large-scale grid-enabled distributed simulation and interoperability

Work in this area focuses on:

1. building Grid-aware distributed simulation engines, addressing issues related to resource management, load balancing, synchronization protocols
2. developing discrete event simulations of large complex systems in different domains
3. Federated simulations and interoperability middleware (such as the High Level Architecture - HLA/RTI).

Distributed algorithms and compact storage for mathematical solutions of large complex systems

This theme focuses on:

1. the modelling and simulation of complex engineering and scientific problems using the full spectrum of numerical and simulation approaches: deterministic, Monte Carlo, linear or non linear, search and optimization, eigenvalue, etc;
2. parallelisation techniques;
3. techniques for efficient data storage and effective distribution.

Data mining and large-scale visualisation

This theme encompasses all work of the proposed center related to Grid-aware data mining and visualisation. Issues addressed include: data cleaning, transportation integration and reduction, languages, mining association rules, classification and prediction techniques, cluster analysis, multidimensional data analysis, and the corresponding visualisation techniques. The University's new HP-supported Spatial and Visualisation Centre plays a crucial role and provides further scope for development in this area.