Chaos Lab

Reference

FAQ

What exactly is chaos?

Deterministic dynamics with sensitive dependence on initial conditions, topological transitivity (dense orbit), and dense periodic points (Devaney's definition). Operationally: a positive Lyapunov exponent in a bounded system.

How is chaos different from randomness?

Chaos is fully deterministic: the same start gives the same trajectory. Random processes have intrinsic stochasticity. Distinguishing the two in observed data uses correlation dimension, surrogate-data tests, BDS test, and the 0-1 test.

Why can't we forecast weather past two weeks?

The atmospheric attractor has a positive Lyapunov exponent of roughly λ₁ ≈ 0.5 day⁻¹, so any initial-condition uncertainty doubles every ~1.4 days. Past two weeks errors saturate at the climatological variance and the forecast becomes statistical.

Is the Mandelbrot set chaotic?

The Mandelbrot set is a parameter set: for each c ∈ M the iteration z ↦ z² + c is bounded but not necessarily chaotic. The boundary of M is where parameters give chaotic dynamics. See Fractal Lab for deep zoom.

What is the Lyapunov exponent?

The asymptotic exponential rate at which infinitesimally close trajectories diverge. λ > 0 means chaos. For maps, λ = lim (1/N) Σ log|f'(x_i)|.

What is the Feigenbaum constant?

δ ≈ 4.6692016091029909..., the limiting ratio of successive intervals between period-doubling bifurcations. Universal across all unimodal maps with a quadratic maximum.

Are strange attractors fractal?

Yes. Strange attractors of dissipative chaotic systems have non-integer Hausdorff dimension; the Lorenz attractor has dim ≈ 2.06, Hénon ≈ 1.26, etc.

How do I detect chaos in data I have?

Embed via Takens, choose τ from average mutual information, m from false nearest neighbours, estimate the largest Lyapunov exponent (Wolf, Rosenstein, Kantz), the correlation dimension, and run BDS + surrogate-data tests. See /measures.

What's hyperchaos?

Chaos with at least two positive Lyapunov exponents. Phase space must be at least 4-D continuous. Examples: hyperchaotic Rössler, hyperchaotic Chen.

What's the difference between chaos and complexity?

Chaos: low-dimensional deterministic dynamics with positive Lyapunov exponent. Complexity: structure emerging from interactions of many simpler parts; may or may not be chaotic. Edge-of-chaos lies between order and chaos and is where most interesting computation happens.

Why does chaos appear in such different systems?

Stretching + folding in phase space produces SDIC + boundedness, which together require nonlinearity. Almost every nontrivial nonlinear system in 3+ continuous dimensions (or 1+ discrete dimension) hosts chaos in some parameter window.

Can chaos be controlled or synchronised?

Yes to both: OGY (1990), Pyragas (1992), Pecora-Carroll (1990). See /control.