A Transactional Architecture for Simulation

Hoverd, Tim; Sampson, Adam T.

doi:10.1109/iceccs.2010.7

Cited by 2 publications

(1 citation statement)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Communication across a network has a considerably higher latency than local communication, but standard distributed programming techniques -such as the use of local proxies [85] and pipelined protocols [86] -can be applied to minimise these overheads, while runtime system integration allows activities to run while others are waiting for network communications [87,88]. CoSMoS has developed design patterns for efficient distributed simulation using messagepassing techniques [89]; ongoing work includes new approaches to temporal synchronisation in distributed simulations [90].…”

Section: Scalable Simulationmentioning

confidence: 99%

Engineering Simulations for Cancer Systems Biology

Bown¹,

Andrews²,

Deeni³

et al. 2012

CDT

View full text Add to dashboard Cite

Abstract:Computer simulation can be used to inform in vivo and in vitro experimentation, enabling rapid, low-cost hypothesis generation and directing experimental design in order to test those hypotheses. In this way, in silico models become a scientific instrument for investigation, and so should be developed to high standards, be carefully calibrated and their findings presented in such that they may be reproduced. Here, we outline a framework that supports developing simulations as scientific instruments, and we select cancer systems biology as an exemplar domain, with a particular focus on cellular signalling models. We consider the challenges of lack of data, incomplete knowledge and modelling in the context of a rapidly changing knowledge base. Our framework comprises a process to clearly separate scientific and engineering concerns in model and simulation development, and an argumentation approach to documenting models for rigorous way of recording assumptions and knowledge gaps. We propose interactive, dynamic visualisation tools to enable the biological community to interact with cellular signalling models directly for experimental design. There is a mismatch in scale between these cellular models and tissue structures that are affected by tumours, and bridging this gap requires substantial computational resource. We present concurrent programming as a technology to link scales without losing important details through model simplification. We discuss the value of combining this technology, interactive visualisation, argumentation and model separation to support development of multi-scale models that represent biologically plausible cells arranged in biologically plausible structures that model cell behaviour, interactions and response to therapeutic interventions.

show abstract