Search

Modelling and simulation of complex systems can create scientific research tools that allow the inaccessible dynamic aspects of systems to be explored in ways that are not possible in live systems. In some scientific contexts, there is a need to be able to create and use such simulations to explore and generate hypotheses alongside conventional laboratory research. The principled complex systems modelling and simulation (CoSMoS) approach was created to support these activities, as a response to a perceived gap in the software engineering development process for simulation. The article presents some of the software engineering motivation for CoSMoS, by exploring this perceived gap. Following from this analysis, the article considers the validation of complex systems simulators, especially where these are to be used in ongoing research. [ABSTRACT FROM AUTHOR]

A collaboration between cancer biologists and academic software engineers has been exploring the development of an agent-based simulator to inform and support work on the dynamics of cell proliferation in the study of prostate disorders. The research has influenced and been informed by the CoSMoS project. This paper presents the simulation project (which is not yet complete). We reflect on the reality of following CoSMoS principles; we describe the domain exploration and show how software modelling approaches (here, Petri nets, state diagrams) can be used to express both biological and software models. We explore fitness for purpose and consider ways to present a fitness argument. We consider issues in choosing simulation media and mapping from domain models through to code. The implementation emphasis is on traceability to support reuse and extension of the simulator, as well as demonstrable fitness for purpose. Initial work on calibration is presented. We discuss the calibration results, that both support and challenge the design and assumptions captured in the domain modelling and development activities. [ABSTRACT FROM AUTHOR]