t = 0.00 s · realizations alive = 16/16
Neutron packet (N) Precursor packet (C) Deterministic reference (RK4)

Stochastic Reactor Kinetics: Neutron Population Noise

The deterministic point-kinetics ODE treats the neutron population as a smooth continuous quantity — useful for control-room instrumentation, but it hides a real physical fact: a nuclear chain reaction is fundamentally a branching random process of discrete, countable neutrons and precursor decays. This 2D simulator solves the identical reaction network (net prompt production/loss, precursor decay, precursor production) with Gillespie tau-leaping, a Monte Carlo method that samples how many times each reaction fires from a Poisson distribution every short time-leap, then updates integer packet counts directly — genuinely distinct mechanics from the RK4 integrator in the 3D version, not a repainted scene. Sixteen independent realizations run side by side so you can watch the statistical spread around the deterministic mean, and — near delayed-critical or barely supercritical reactivity — watch some realizations randomly die out entirely while others take off, the real branching-process hazard behind reactor startup and subcritical-multiplication measurements.