Simulation Lab

An SIRS epidemic on a random contact network, from the Python ABM lab. Infected agents lose health and can die, recovered agents stay immune for a while and then become susceptible again, and a small chance of infection from outside keeps the disease circulating. Susceptible agents watch their neighbours and isolate when too many are sick. Optuna searched for an infection that never dies out but returns in regular waves without wiping out the population; this is the winning setting, run offline and played back.

Engine

Mesa 3.5 · NetworkX

Network

280 nodes · average degree 4 · 10 infected at start

Key parameters

spread 0.59 · immunity wanes 0.074 · outside infection 0.029

Run

500 steps · seed 33700

Tuning

Optuna, 500 trials · winner #337

Score

0.8931 of 1 (persistence, waves, regularity, survival)

Colours

white = susceptible · red = infected · green = recovered · dim = empty