Research areas of the groups involved
We are drawing together:
- Applications of modelling: communications networks (Electronic Engineering); cognitive science and software systems (Computer Science); fluid flows and particle systems (Engineering).
- Foundations of modelling: computational modelling (Engineering): mathematics of modelling (Mathematics) : statistical approaches (Mathematics); and visualisation and perception (Computer Science).
Electronic Engineering
Networks: Within the UK, current research in the Networks Group is unique in its approach to the challenges of very large-scale systems. Funded by EPSRC the group is evaluating the use of high-performance computing to address benchmarking, representation and reproducibility (particularly with respect to simulation) for networks of enormous scale. This work is the first time UK supercomputers have been used for this.
The size and complexity of networked communications systems has created very significant challenges for any science of networking. Particularly important are reproducibility and benchmarking. Reproducibility implies system validation; benchmarking requires both validation and correct modelling.
Other projects in the Networks Group are tackling methodological challenges, e.g. the interactions between topological and traffic properties in large topologies. A pilot multidisciplinary project studies how statistical techniques can bound measurement errors across large networks.
Mathematical Sciences
Statistics - Design of Experiments: QMUL is world leading in the Design Of Experiments (DOE) and surveys - a large proportion of the UK’s experts are based at QMUL, including Gilmour, Bailey, Bogacka and Coad.
While DoE has been widely applied to research in the Biological Sciences and many aspects of Industrial Engineering, its potential for optimising
numerical experiments (effectively simulations) of complex systems has not yet been fully realised. Partly this is due to the fact that many of these (man-made / physical / logical) systems have not yet been thought of as being open to study through optimally designed experimentation.
Combinatorics: QMUL is a home to the renowned group of combinatorialists led by Prof. Peter Cameron. One of the research areas of the group are combinatorial aspects of the design of experiments ( http://www.maths.qmul.ac.uk/~pjc/design/drg.html ) .
Dynamical Systems: This group (headed by Prof David Arrowsmith) has immense strength in depth, its researchers covering many areas that are of potential importance in this project, in particular: nonlinear dynamics and statistical physics, e.g. chaotic behaviour in Hamiltonian systems, bifurcation theory, complex maps, distributed systems, control systems, applications of chaotic systems to packet traffic, and algebraic dynamics. As well as the EU grant on the dynamics of the network of networks, the Group is part of the ESF network STOCHDYN (Stochastic Dynamics: Fundamentals and Applications) together with various other groups from eleven European countries, which aims at enhancing research and collaboration in the field of dynamical systems, stochastic processes, and non-equilibrium statistical mechanics.
Computer Science
Logic and semantics: QMUL has a breadth and depth in logic and semantics that is almost unmatched in the UK. The Group’s work has spearheaded several developments, including separation logic, logic for continuous systems, information theory for security, process types for web services.
This Group represents systems quite unlike physical systems of fluid flows, or even the flow of data through communications networks. Here research is concerned with systems that are immense logical structures, and the particular challenges focus on validation and reliability. There have already been some suggestions that there may be significant value in an interchange of ideas between researchers in topological and logical validation.
Biologically inspired systems: Understanding the computational feat of biological vision is a major scientific challenge. The pattern of light falling on the human retina or camera sensor contains a vast amount of system information, all of which is implicit apart from the image brightness recorded for each of the vast number of rods, cones or pixels. It has been recognised since the 1970’s that the key role of a biological or computer vision system is to extract the information in the image and make it explicit by describing it within a system that can represent the information we want to extract. To a considerable extent theory in biological vision has been guided by explorations in computer vision over the intervening period.
Engineering
Fluid flows: Research in the field of turbulent fluid mechanics uses both computational and experimental methods, and requires the use of parallel computing systems to investigate complexity and topology in large-scale turbulent flows. This field has traditionally used direct numerical simulation, including boundary layers and jets. Much of this work seeks to reduce the information content of large-scale, complex flows to a set of criteria that can be used to predict future behaviour. Experimental measurements of turbulent flows typically produce thousands of data streams that require intensive processing to reveal the underlying modes of activity and the evolution of flow patterns. Given the need to develop parsimonious models, this is clearly analogous to some of the work in the Networks Group (Electronic Engineering – Networks), as well as the biologically inspired work of Prof. McOwan.
Ultra large-scale particle simulations: The other main research area of interest for this project in Engineering is that of computational mechanics, particularly the simulation of systems of discontinua comprising potentially billions of particles, anachievement that would result in unforeseen fundamental research achievements in systems from nano-scale to terrestrial bodies. The theme is led by Prof Ante Munjiza, who was the first to model a one million particle system and wrote the first textbook on FEM/DEM. He currently heads a major EPSRC project in collaboration with major mining companies to apply these techniques in geoscience.