Gaussian process emulation of an individual-based model simulation of microbial communities

• DOI: 10.1016/j.jocs.2017.08.006
• Publication Date: 2017-08-15T00:31:05Z
• Authors: O.K. Oyebamiji; D.J. Wilkinson; P.G. Jayathilake; T.P. Curtis; S.P. Rushton; B. Li; P. Gupta
• Keywords: Microscale chemistry; Simulation Modeling

The ability to make credible simulations of open engineered biological systems is an important step towards the application scientific knowledge solve real-world problems in this challenging, complex engineering domain. An type design and management wastewater treatment systems. One crucial aspects biology approach study run a simulation communities. However, challenging because they often involve large number bacteria that ranges from order 1012 (a baby's microbiome) 1018 plant) individual particles, are physically complex. Since models computationally expensive, due computing constraints, consideration only limited set scenarios possible. A simplified problem use statistical approximation ensembles derived at fine scale which will help reducing computational burden. Our aim paper build cheaper surrogate individual-based (IB) model microbial focuses on how emulator as effective tool for studying incorporating microscale processes efficient way into macroscale models. main issue we address strategy emulating high-level summaries IB data. We Gaussian process regression emulation. Under cross-validation, percentage variance explained univariate 83–99% 87–99% multivariate emulators, both biofilms floc. emulators show approximately 220-fold increase efficiency. sensitivity analyses indicated substrate nutrient concentration nitrate, carbon, nitrite oxygen well maximum growth rate heterotrophic most parameters predictions. observe performance single depends hugely initial conditions sample size taken normal approximation. believe development strategic importance using understanding enable solving.