Applications › Agent-based modelling of emergent behaviour
Applications · Complex systems
Agent-based modelling of emergent behaviour
Systems whose behaviour cannot be derived from their components — markets, epidemics, grids, cities — modelled by simulation rather than by solution.
The problem
Some systems have no closed form. The behaviour that matters — a cascade, a bubble, a tipping point — emerges from many local interactions and is not visible in any individual component. For these, simulation is not a second-best approach to analysis; it is the only approach.
Agent-based modelling is the standard tool, and its weakness is well known: the parameter space is enormous, calibration is hard, and it is easy to produce a model that reproduces history and predicts nothing. That calibration problem is itself an optimisation problem, and that is where the rest of the centre becomes useful.
Agent-based simulation with quantum-inspired components; calibration cast as global optimisation and handed to annealing or hybrid solvers; and analysis of emergent structure using fractal and complexity measures.
Established. Agent-based modelling is a mature discipline with good tooling. The contribution here is in the calibration and parameter-search layer, and in the analysis of the output.
Active. The Quantum-inspired Agent-based Simulation Laboratory is working in this area; the foundations group supplies the complexity and fractal analysis.
Economic and financial system modelling; epidemic and public-health scenarios; urban and infrastructure planning; energy market behaviour.
Behavioural data or plausible behavioural rules; a defined question the model is meant to answer — general-purpose models answer nothing in particular; and compute for the parameter sweep, which is usually the largest cost.
Talk to us about this
Complex-systems work goes wrong when the question is vague. Bring a decision you need to make, and we will tell you whether a model can inform it.
Last updated 14 September 2026.