Lorenz Attractor
The butterfly effect visualized in 3D phase space
Edward Lorenz and Chaos Theory
In 1961, MIT meteorologist Edward Lorenz was running numerical weather simulations on an early computer. To save time, he restarted a simulation from the middle, entering data rounded to three decimal places instead of the full six. The result diverged completely from the original — weather forecasts became useless after just a few days.
This discovery led Lorenz to develop a simplified model of atmospheric convection in 1963, producing the now-famous Lorenz Attractor. His 1972 talk "Predictability: Does the Flap of a Butterfly's Wings in Brazil Set Off a Tornado in Texas?" popularized the term "butterfly effect" to describe sensitive dependence on initial conditions.
The Lorenz Equations
The Lorenz system consists of three coupled differential equations:
dx/dt = σ(y − x)
dy/dt = x(ρ − z) − y
dz/dt = xy − βz
Where:
- σ (sigma) = Prandtl number (ratio of momentum to thermal diffusivity)
- ρ (rho) = Rayleigh number (temperature difference)
- β (beta) = geometric factor
Classic values: σ=10, ρ=28, β=8/3. These produce the iconic butterfly-shaped attractor.
Strange Attractors and Phase Space
The Lorenz Attractor is a strange attractor — a set in phase space with fractal structure. Unlike fixed points or limit cycles, strange attractors:
- Never repeat — trajectories never pass through the same point twice
- Remain bounded — all solutions stay within a finite region
- Have fractal dimension — approximately 2.06 for the Lorenz attractor
- Exhibit sensitive dependence — nearby trajectories diverge exponentially
The system alternates chaotically between circling the left and right lobes of the butterfly, with the choice unpredictable despite being deterministic.
Lyapunov Exponents and Predictability
The Lorenz system has a positive Lyapunov exponent, quantifying exponential divergence. If two trajectories start $\epsilon$ apart, after time $t$ they are approximately $\epsilon e^{\lambda t}$ apart, where $\lambda \approx 0.9$ for standard parameters.
This means:
- Doubling time ≈ 0.77 time units
- After 10 time units, trajectories 1000× farther apart
- Finite precision (e.g., 10 decimal places) limits prediction to ~23 time units
This is why weather forecasts become unreliable beyond ~10 days, despite our equations being deterministic.
The Lorenz System: Where Chaos Theory Began
In 1963, meteorologist Edward Lorenz discovered that a simple 3-equation model of atmospheric convection produced unpredictable behaviour. Three coupled differential equations describe the system:
dx/dt = σ(y−x), dy/dt = x(ρ−z)−y, dz/dt = xy−βz
With classic parameters σ=10, ρ=28, β=8/3, the system traces the iconic butterfly-shaped strange attractor.
What is a Strange Attractor?
An attractor is a set of states towards which a system evolves. A strange attractor has fractal structure — it occupies a non-integer dimension (the Lorenz attractor has dimension ≈2.06). The trajectory never repeats exactly, though it stays confined to the butterfly shape forever. This is deterministic chaos: bounded, structured, yet never periodic.
Parameter Exploration
- σ (Prandtl number): ratio of viscous to thermal diffusion
- ρ (Rayleigh number): drives convection intensity — below ρ≈24.74, the system converges to fixed points; above, chaos emerges
- β (geometric factor): relates to cell dimensions
Adjusting these parameters transitions the system between stable fixed points, periodic orbits, and chaotic behaviour.
Chaos Theory and Weather Prediction
Lorenz's discovery explains why weather cannot be predicted beyond approximately 2 weeks. Any measurement uncertainty — no matter how tiny — grows exponentially, making long-term prediction impossible. "Predictability: Does the Flap of a Butterfly's Wings in Brazil Set Off a Tornado in Texas?" — Lorenz's 1972 lecture title that coined the term "butterfly effect."
Experiments to Try
- Set ρ=28 for classic chaotic behaviour
- Reduce ρ below 24 and watch the system settle into fixed points
- Try ρ≈99.65 for a periodic orbit within chaos
- Compare two trajectories with slightly different starting points to observe sensitive dependence
Key Equations
| Concept | Formula | Notes |
|---|---|---|
| Lorenz system (x) | dx/dt = σ(y − x) | σ: Prandtl number (default 10) |
| Lorenz system (y) | dy/dt = x(ρ − z) − y | ρ: Rayleigh number (default 28) |
| Lorenz system (z) | dz/dt = xy − βz | β: geometric factor (default 8/3) |
| Chaos onset | ρ > ρc ≈ 24.74 | Below: fixed points; above: chaos |
| Lyapunov exponent | λ1 ≈ 0.906 | Positive ⇒ sensitive dependence on ICs |
| Attractor dimension | dH ≈ 2.06 | Fractal: non-integer Hausdorff dimension |
Curriculum Relevance
| Level | Topic | Relevance |
|---|---|---|
| A-Level | Differential equations | Systems of ODEs, numerical integration (Euler/RK4) |
| IB / AP | Calculus, mathematical modelling | Phase portraits, parameter sensitivity |
| Undergraduate | Dynamical systems, fluid mechanics | Navier-Stokes approximation, chaos theory |
| Postgraduate | Nonlinear dynamics, ergodic theory | Strange attractors, Lyapunov spectra, fractal dimensions |
Share this simulation
Send this page to students or colleagues.
Related simulations
Continue with similar experiments.
📖 Recommended Reading
Deepen your understanding with our in-depth articles.
🔒 Unlock All 32 Simulations
Get unlimited access to all 32 simulations — including the Lorenz Attractor, Logistic Map, Mandelbrot Set, and more with MySimulator Premium.