Processing ever-larger volumes of data raises new challengess in the development and use of predictive models of complex systems of interacting elements.
The development and use of predictive models is a common strategy in the analysis of complex systems of interacting elements. In some cases, the primary task is to model from observed data, building a model which best encapsulates the current knowledge of the system so as to aid our understanding. Such models arise across the natural sciences, the life sciences and, increasingly, the social sciences. Examples include social communities of insects, gene expression networks, networks of traffic flow, biochemical pathways and populations of interacting nerve cells. Where the system under study is a designed or engineered one, the principal aim is to construct a model which abstracts from detail, but which will nevertheless facilitate reasoning and analysis. In the context of computer systems, the development and analysis of models complement traditional design review and testing. Artefacts built in software and hardware must often be scrutinised by appropriate models to ensure the appropriateness of their behaviour. This is increasingly seen as crucial if complex computer systems - on which so much of the modern hi-tech economy depends - are to function reliably, safely and efficiently.
Computational modelling is now an established approach to understanding the nervous system, identified as one of the UKCRC Grand Challenges in Computing. Relevant work in Scotland therefore ranges from foundational modeling, through neuroscience and systems biology. A particular focus of our proposed work will be scalable analysis, not just because very complex systems of systems are our targets, but because the models themselves are so vast that they require new analysis techniques, system description languages, methods, and tools. Scottish groups are world-leading experts in aspects of this theme. They already collaborate on the application of process algebras to systems biology, and on model checking techniques. By bringing this work together, and by also stimulating neuroscience modeling and visualization work, SICSA will enable Scotland to take a leading international role in developing scalable analysis.
All past Modelling and Abstraction events are available to view here.