Classical Chaos
Newton's laws are deterministic: an initial state fixes a trajectory. Yet some systems become unpredictable in practice. Classical chaos explains this apparent paradox. It does not add randomness to the equations; it rapidly amplifies small uncertainties. This lesson connects phase space, Hamilton's equations, and the emergence of complex behavior.
Determinism and predictability
A system is deterministic if its equations and initial state determine its evolution. It is predictable only while initial uncertainties remain sufficiently small. A chaotic system can be deterministic and quickly lose predictability.
Sensitivity to Initial Conditions
Consider two initial states separated by a small distance $\delta(0)$ in phase space. In a chaotic region, their separation grows approximately as:
The number $\lambda$ is a Lyapunov exponent. A positive largest exponent signals exponential divergence of nearby trajectories. The Lyapunov time $t_L=1/\lambda$ is the timescale over which an error grows by a factor $e$.
Prediction Horizon
Let the initial uncertainty be $\delta_0$ and the largest acceptable error be $\Delta$.
Set $\delta_0e^{\lambda t_*}=\Delta$.
The useful prediction time is $t_*=\lambda^{-1}\ln(\Delta/\delta_0)$. Improving initial precision by a factor of one thousand adds only a logarithmic amount of time.
Why Nonlinearity Matters
A linear system simply superposes its solutions and cannot produce the repeated folding needed for bounded chaos. Nonlinear terms couple degrees of freedom. They stretch phase-space regions and fold them back into a finite domain. Stretching creates sensitivity; folding mixes trajectories.
Poincaré Sections
A Hamiltonian trajectory with two degrees of freedom lives in a four-dimensional phase space. A Poincaré section records successive intersections with a chosen surface. It turns a continuous flow into a readable discrete map:
- a periodic orbit produces finitely many points;
- quasi-periodic motion draws a closed curve;
- a chaotic region fills an irregular area of points.
Example: The Driven Pendulum
A damped pendulum driven by a periodic force obeys:
The $\sin\theta$ term makes the equation nonlinear. Depending on $A$, $\Omega$, and $\gamma$, the pendulum can oscillate periodically, undergo period doubling, and become chaotic. Chaos therefore needs no large collection of objects: one angle, its velocity, and periodic forcing are enough.
Order Within Chaos
Chaos does not mean absence of structure. Hamiltonian systems preserve phase-space volume by Liouville's theorem. Regular islands can coexist with chaotic seas. The KAM theorem says that some quasi-periodic trajectories survive a small perturbation of an integrable system, while resonances break first.
Chaos and randomness
Classical chaos amplifies ignorance about an initial state. Quantum randomness, in the standard interpretation, concerns probabilities of measurement outcomes. Both limit prediction, but they are not the same mechanism.
Exercises
- If $\lambda=0.5\,\mathrm{s}^{-1}$, how long does it take an error to grow by a factor of 100?
- Explain why a chaotic trajectory does not violate energy conservation in an autonomous Hamiltonian system.
- Compare the Poincaré sections of periodic and chaotic motion.
- A chaotic system can be deterministic while remaining unpredictable at long times.
- A positive Lyapunov exponent measures exponential amplification of initial errors.
- Nonlinearity enables stretching and folding of trajectories in phase space.
- Poincaré sections distinguish periodic, quasi-periodic, and chaotic motion.
- Chaos has geometric structure and still obeys the system's conservation laws.