[Zurück]

G. Schneckenreither, G. Zauner, N. Popper, F. Breitenecker:
"Implementing a Large-Scale Spatial Epidemic Model as Agent-Based System Using GPGPU and Databases";
Vortrag: EUROSIM 2010 - 7th Congress on Modelling and Simulation, Prague, Czech Republic; 06.09.2010 - 10.09.2010; in: "Proceedings of the 7th Congress on Modelling and Simulation", M. Snorek, Z. Buk, M. Cepek, J. Drchal (Hrg.); Vol.2 Full Papers (2010), ISBN: 978-80-01-04589-3; S. 868 - 874.

Epidemiological Models are basically composed of medical, demographic and spatial features of arbitrary complexity depending on available data and purpose of the simulation. With increasingly subtle structures the formulation and implementation becomes more and more challenging. Additionally the performance limit of simulators, applications and hardware interferes with the models fidelity. The purpose of this investigation is to find and test new approaches for implementing and simulating systems with large data volume and complexity. Some approved approaches/concepts involve matrix based programming languages, parallelisation, cellular automata and agent-based systems. Our goals are to swap costly calculations to equivalent methods with lower time penalty, make use of fundamental concepts such as databases during simulation or to facilitate access to advanced programming techniques like GPGPU (computing on graphics devices). This contribution presents and evaluates different approaches towards implementing large spatial systems using cellular automata and agent-based approaches. The discussion is based on a spatial model for simulating the spread of two competing diseases (serotypic shift).

Cellular Automaton, Spatial Agent-Based System, Database, GPGPU, Disease Propagation