Navigation menu

Simulation of a wound tool punching a hole in the blood vessel wall. Chemicals (shown in green) are released, causing platelets to become sticky. Some small clots can be seen being carried away downstream.

A little later after the wound tool removed: more clots have formed and are flowing away. Some larger clots have formed.

Authored by Fred Barnes
Published 7 June 2007

The University of Kent, in collaboration with the University of York, has been awarded £398,058 by EPSRC to investigate and develop tools for the simulation of complex systems -- named "CoSMoS" (complex systems modelling and simulation). The combined funding for the project, which includes £144,000 from Microsoft, is approximately £1.1M. The project starts in October 2007 and will run over four years.

The project aims to provide tools to allow researchers to study, model and simulate complex systems, involving millions of individual interacting components. A classic example, which was studied as part of the earlier "TUNA" project (an EPSRC funded feasibility study), is the simulation of blood-clotting. For a given section of a blood vessel, involving millions of platelets and various chemical factors, we can simulate the behaviour when the vessel wall is damaged -- which ultimately leads to a clot (composed of 'sticky' platelets) forming around the wound. Such simulations allow some level of biological experimentation without human or animal participation (where there may be ethical issues). Observed behaviour from the real world is used to improve the underlying models, which are then simulated and compared to expected results. Key to the research is an idea of emergent behaviour: we do not program the platelets to build clots around wounds -- that arises as a result of complex interactions between platelets, chemical factors and the environment.

Professor Peter Welch and Dr. Fred Barnes have significant experience and expertise in the creation of highly concurrent software systems using the occam-pi programming language, involving millions of concurrent processes. Such levels of concurrency are otherwise unattainable, typically limited to hundreds or thousands of processes in existing concurrency systems. The occam-pi language will be used to realise the simulation portion of the work at Kent, whilst researchers at York concentrate on the modelling, and simulation in other areas (direct hardware parallelism using FPGAs).