Chaos and nonlinear dynamics

Sensitive dependence, bifurcations, strange attractors, Lyapunov exponents.

A deterministic system is one whose future is fixed by its present: give the equations and the initial state, and the trajectory is settled for all time. For two centuries that was taken to mean predictable — Laplace’s demon, knowing every position and velocity, could read off the future like a table. Chaos is the discovery that determinism and predictability are not the same thing. A perfectly deterministic system with as few as three variables, and no randomness anywhere in its equations, can be unpredictable in practice over any but the shortest horizon, because it amplifies the tiniest uncertainty in its initial state exponentially fast.

The mechanism is sensitive dependence on initial conditions: nearby trajectories separate at an exponential rate, so a state known to one part in a billion becomes wholly uncertain after a fixed, often short, time. The motion is nonetheless bounded and structured — it lives on a strange attractor, a fractal set the trajectory winds around forever without repeating. And the route a system takes from orderly to chaotic behaviour, as a parameter is turned, is shared across systems that have nothing physically in common: the period-doubling cascade of a population model and of a dripping tap accumulate at the same universal rate.

This chapter develops the working subset, from the simplest possible example to the canonical one.

The chapter leans on the phase-plane picture from the ODE chapter and the numerical ODE solvers used to integrate the flows; chaos is where the nonlinear terms those chapters set aside finally take centre stage.