Deploying HPC to biofilm simulations
by Denis Taniguchi
16:00 (40 min) in USB 2.022
Biofilm simulation can be a powerful tool to provide new ways to improve the understanding of microbial communities. The impact of biofilms can be observed in many areas such as wastewater treatment, pipe corrosion, and medical devices. The Newcastle University Frontiers in Engineering Biology(NUFEB) is a project devoted to shed new light into wastewater treatment, and possibly beyond, by the use of biofilm simulation. It employs an individual-based model (IbM) that represents each bacteria as a separate entity capable of growing, dividing, decaying and eventually dying. The physical, biological and chemical processes are modelled using the first principles endowing the model with enough precision to replicate complex phenomena (for more details see Jayathilake2017).
Scale is the greatest challenge for those seeking to use IbM simulations to create real-world impact. To allow experimental validation of simulations for wastewater treatment, due to the diversity of the colonies and the convoluted nature of their dynamics, bigger and longer simulations are required. Therefore the use of HPC is essential.
In this talk I will delve into the techniques employed in the NUFEB project to enhance the computational performance of the simulator. I will describe the problems faced in the parallelisation of the code, the computational bottlenecks and solutions employed to overcome them (or sometimes live with them), and on-going efforts and future plans which include the use of performance portability (use of multiple computing devices, such as GPUs).