Speakers
Description
GreenPhysECS is an exploratory project investigating whether the Entity-Component-System (ECS) architectural model can make parallel, energy-aware research software more accessible to early-career researchers. ECS - widely used in the games industry - has seen little application in research. This project examines whether adopting ECS from the outset helps novice developers produce concurrent-friendly simulation code with integrated energy tracking.
This talk presents student-led case studies from MPhys projects across two distinct domains. The first applies ECS to computational fluid dynamics, implementing both finite difference and smoothed particle hydrodynamics approaches to simulate compressible gas flow, with CPU energy consumption tracked across different time integration methods. The second applies the same ECS framework to stochastic epidemic modelling, progressing from SIR (Suspected-Infected-Recovered) models to spatial and network extensions and finally to a HUD (Healthy-Undetectable-Detectable) model for early detection of crop disease using sentinel plants, including exploratory extensions beyond the initial models considered.
Taken together, these case studies provide early evidence that ECS can support the development of modular, naturally parallelisable simulation code by early-career researchers, while making energy measurement and comparison straightforward to integrate into the development process.