CoSMoS: Complex Systems Modelling and Simulation
Lead Research Organisation:
University of Kent
Department Name: Sch of Computing
Abstract
Our proposal builds capacity in generic modelling tools and simulation techniques for complex systems, to support the modelling, analysis and prediction of complex systems, and to help design and validate complex systems. Drawing on our state-of-the-art expertise in many aspects of computer systems engineering, we will develop CoSMoS, a modelling and simulation process and infrastructure specifically designed to allow complex systems to be explored, analysed, and designed within a uniform framework.
People |
ORCID iD |
| P Welch (Principal Investigator) | |
| Frederick Barnes (Co-Investigator) |
Publications
A.T.Sampson
(2008)
Investigating Patterns for the Process-Oriented Modelling and Simulation of Space in Complex Systems
in Artificial Life XI: Proceedings of the Eleventh International Conference on the Simulation and Synthesis of Living Systems
Bown James
(2012)
Engineering Simulations for Cancer Systems Biology
in CURRENT DRUG TARGETS
Hoverd T
(2010)
A Transactional Architecture for Simulation
P.H.Welch
(2013)
Mutually Assured Destruction (or the Joy of Sync)
in Communicating Process Architectures
P.H.Welch
(2013)
Life of occam-Pi
in Communicating Process Architectures
P.H.Welch
(2013)
Communicating Process Architectures 2013: Proceedings of the 35th. WoTUG Technical Meeting
in Communicating Process Architectures
Pedersen J
(2018)
The symbiosis of concurrency and verification: teaching and case studies
in Formal Aspects of Computing
Polack F
(2020)
On Developing and Validating Dynamic Systems: Simulation Engineering.
in The Journal of Object Technology
Polack F
(2010)
Reflections on the Simulation of Complex Systems for Science
| Description | Computer-based simulation is a key tool in many fields of scientific research. In-silico experiments can be used to explore and understand complex processes, to guide and complement in vitro and in vivo experiments, to suggest new hypotheses to investigate, and to predict results where experiments are infeasible. Simulation is an attractive, accessible tool: producing new simulations of simple systems is relatively easy. But it is also a dangerous one: simulations are often complex, buggy, and difficult to relate to the real-world system. On the CoSMoS project, we have produced a principled and rigorous approach to developing and validating Simulations as Scientific Instruments. The approach is non-proprietary, and development language-neutral, and can be used in conjunction with in-house specific languages and technologies. It emphasises communication between the domain expert (end user) and the simulation developers, through cooperative building of a set of well-defined models. This ensures that the resulting simulation is fit-for-purpose, justifiably valid, and open. The approach has been tested on a range of real-world case studies, mostly from the immuno-biology domain, but also in some socio-technical areas. The minimal process is documented in a 40 page technical report (available from the project website). A user handbook is being written (to be published by Springer), adding discussion of patterns, calibration, sensitivity analysis, and validation issues. The CoSMoS approach has been developed on, and validated through, a range of real world case studies, in partnership with domain experts. These studies include models and simulations for lymphocyte rolling and hepatic granuloma formation (with the York Centre for Immunology and Infection), EAE (Experimental autoimmune encephalo-myelitis, a mouse proxy for Multiple Sclerosis, with the Torrey Pines Institute for Molecular Studies, USA), prostate cell division and differentiation (with the York Cancer Research Unit), and a simulation of a multiple dwelling thermodynamic model, with the European Institute for Energy Research (EIFER, Karlsruhe). The CoSMoS team has developed various models and meta-models, and structured validation arguments, that underlie high-quality agent-based simulations. All modelling is done in close partnership with relevant domain experts, to ensure that the resulting system is fit-for purpose, by capturing and interrogating the assumptions, approximations, and rationale underlying its development. Validation of modelling, implementation, and experimentation is addressed through structured argumentation techniques adapted from the safety critical arena. The CoSMoS team has developed language-neutral techniques for implementing large-scale complex systems simulations using multicore and cluster computing resources. This work includes scalable approaches to simulating spatial interactions between agents, high-performance runtime systems that schedule simulation work automatically and dynamically across multicore CPUs (possibly the fastest and most effective currently available), and techniques for efficient network communication within cluster simulations. This lightweight parallel programming paradigm has been realised in the parallel programming language occam-pi (supporting millions, and more, parallel processes), as a Java class library (supporting thousands of parallel processes) and as C and C++ libraries. CoSMoS simulations can achieve near-linear speedup on medium-sized clusters. This work has also driven novel parallel language developments (such as session protocols for occam-pi and language extensions to support formal verification of concurrent systems). The paradigm, used for many CoSMoS case studies, is especially well suited for agent-based simulation, allowing agents to execute naturally as parallel processes, and for research into the safe design and management of systems with complex emergent properties. |
| Exploitation Route | Any field that requires (computer) modelling of complex systems benefits from the methods, frameworks and demonstrators produced by this project. We have provided a unified modelling and simulation framework to build, analyse, and compare different design and implementation choices. The Open Source simulation framework enables new simulations, and extensions and modifications to the framework (e.g. new capabilities, new platforms) to be built by external groups. We have extended parallel language and (scalable multicore) runtime kernels, giving new options for efficient implementation of such models. Over 40 refereed publications and books document these outcomes. |
| Sectors | Aerospace, Defence and Marine,Chemicals,Communities and Social Services/Policy,Digital/Communication/Information Technologies (including Software),Education,Pharmaceuticals and Medical Biotechnology,Other |
| URL | http://www.cosmos-research.org/ |
| Title | Kent Retargettable occam Compiler (KRoC) |
| Description | Compiler and multicore run-time system for the occam-pi concurrent programming language. |
| Type Of Technology | Software |
| Year Produced | 2012 |
| Open Source License? | Yes |
| Impact | KRoC provides a demonstrator providing proof-of-concept technology for performance scalability (w.r.t. the number of cores) of complex concurrent systems (including real-time and complex modelling). Although occam-pi is (regrettably) not a `mainstream' system language, the mechanisms developed for its implementation -- that provide efficient and multicore scalability -- and for the verification of the concurrent behaviour of its programmed systems provide targets for mainstream technologies to achieve. Published papers (listed as outcomes from our EPSRC `CoSMoS' research grant) explain the mechanisms developed for this. |
| URL | http://www.cs.kent.ac.uk/projects/ofa/kroc/ |