Many complex systems can be modelled realistically using an agent-based approach. In a number of application areas this is becoming more popular and is leading to significant progress in understanding how these systems work. One of the drawbacks with much of the work in this area is that the models are essentially small-scale. The Flexible Large-scale Agent-based Modelling Environment (FLAME) has been developed for use on parallel supercomputers and can thus handle models where there may be millions of agents of different types. Examples of the application of the framework to problems in biology – at the molecular, cellular and ecological levels will be discussed. The framework is also being used to model social and economic systems with a view to informing policy decisions at national and regional levels. Some examples will be given of these models. Finally, we consider the reliability and integrity of these models and look at some techniques for formally specifying, validating and testing large-scale agent-based models.
Professor Mike Holcombe
Mike Holcombe, Professor of Computer Science, University of Sheffield.
Member of Verification and Testing and of Computational Biology Research groups. Director of the Sheffield Software Engineering Observatory. Member of Board of the software company, epiGenesys.
Current research interests: software testing, empirical software engineering, computational biology, computational economics.
Current teaching interests: practical software engineering and agile development, enterprise in computing.