Exascale Agent-based Modelling for PoLicy Evaluation in Real-time (ExAMPLER)

Imagine what you could do with agent-based modelling if you had a billion laptops at your disposal, instead of just one (desktop computing) or a thousand (cluster computing). That kind of computing power has the potential to transform agent-based modelling, provided we can put together appropriate software, data and institutional support.

This is what the ExAMPLER project will explore, and more...

Exascale computing exploits GPU (Graphical Processing Unit) technology to generate calculation speeds of 1018 floating point operations per second. A laptop does about 109. Agent-based modelling is a form of computer simulation that gained prominence during the Covid crisis to evaluate policy options. However, it is still not typically making use of high-performance computing resources, sometimes because these are not set up or managed to support our needs. So, ExAMPLER will:

  • Explore the state-of-the-art in high-end computing use by the empirical agent-based modelling community;
  • Benchmark that against other disciplines;
  • Co-construct visions of future agent-based modelling underpinned by exascale computing;
  • Develop a roadmap to realize those visions.
ExAMPLER team photo (from left to right: Matt Hare, Gary Polhill, Mike Batty, Doug Salt, Alison Heppenstall, Richard Milton, Ric Colasanti)