Back

Lorenz attractor

Change parameters to explore chaotic trajectories and sensitivity to initial conditions.

Presets
Parameters
σ (sigma) — Prandtl number
ρ (rho) — Rayleigh number
The stability boundary depends on σ and β. Check the regime banner above.
β (beta) — aspect ratio
Speed and display
Drawing speed
Trail length
Projection
Controls
Live values (trajectory A)
0.00
x (current)
0.00
y (current)
0.00
z (current)
—
Current wing
0.0e+0
Separation of the two trajectories Δ
—
λ (finite-time estimate)
0.0
Elapsed time t
Chaotic region: the trajectory never repeats the same path, and the two trajectories diverge exponentially.
Butterfly attractor (divergence of two trajectories)

3D: drag to rotate, two-finger pinch to zoom | wheel to zoom | Space to play/pause

Trajectory A (post-preparation reference point) Trajectory B (x₀ + 10⁻⁵) Current position
Same equations, slightly different initial states: Blue and red start Δx₀=10⁻⁵ apart. In the Classic chaos setting they initially look like one path, then follow different routes. Separation Δ stretches and contracts; it does not grow uniformly forever.
Theory & Key Formulas
$$\frac{dx}{dt}= \sigma(y - x)$$ $$\frac{dy}{dt}= x(\rho - z) - y$$ $$\frac{dz}{dt} = xy - \beta z$$

For nearby trajectories, $\Delta(t)\approx\Delta_0 e^{\lambda t}$ describes local growth, not separation after it reaches the attractor scale. The displayed λ is a finite-time estimate, not a stand-alone proof of chaos or periodic motion.

Numerical integration: 4th-order Runge–Kutta (RK4, dt = 0.005)

What this calculator covers

Tune σ, ρ, β in real time. Two trajectories always run from nearly identical initial conditions (Δx₀=10⁻⁵); watch their separation grow — the butterfly effect made visible.

Formulas and key points
Quick answer
The Lorenz equations are dx/dt = σ(y − x), dy/dt = x(ρ − z) − y, dz/dt = xy − βz. This tool integrates them with 4th-order Runge–Kutta (dt = 0.005) and visualizes sensitive dependence on initial conditions — the butterfly effect.

Preset β=2.667 is rounded from 8/3. The solver uses the displayed input values. λ is a finite-time estimate along the displayed trajectory. Its short-time sign is not a classification.

Methods and comparison values: Lorenz (1963) · Sprott: Lyapunov exponent

What is the Lorenz Attractor?

Explanation in dialogue
🙋
What exactly is the "butterfly effect" I keep hearing about, and how does this simulator show it?
🎓
The simulator always compares two trajectories, starting Δx₀=10⁻⁵ apart after a preparation interval. In the Classic chaos setting they can initially look identical and later follow different routes. There is no trajectory-count control: compare their separation Δ and coordinates while changing one parameter at a time.
🙋
Wait, really? So those three equations control everything? What do the σ (sigma) and ρ (rho) sliders actually do?
🎓
Yes, those three equations are the heart of it. The sliders change the system's behavior. For instance, σ is related to fluid viscosity and thermal conductivity. The famous chaotic "butterfly" shape appears when ρ is around 28. Try moving the ρ slider down to 10—the system becomes stable and spirals into a fixed point. Now push it up to 28 and see chaos emerge!
🙋
That's cool. But why is this relevant to engineering? It just looks like a pretty 3D spiral.
🎓
RK4 approximates the time evolution of this educational model. Engineering solvers select integration methods, error controls and validation appropriate to their own equations. This Lorenz visualization illustrates sensitivity to initial conditions; it does not calculate aircraft flutter, structural safety or a weather forecast.

Physical Model & Key Equations

The Lorenz system is a set of three coupled, nonlinear, ordinary differential equations. They were derived as a simplified model of atmospheric convection (heat transfer). The equations govern how the state variables—x (convection intensity), y (temperature difference), and z (vertical temperature profile)—change over time.

$$\frac{dx}{dt}= \sigma(y - x)$$ $$\frac{dy}{dt}= x(\rho - z) - y$$ $$\frac{dz}{dt} = xy - \beta z$$

Variables: $x, y, z$ are the system states. Parameters: $\sigma$ (Prandtl number, ratio of momentum to thermal diffusivity), $\rho$ (Rayleigh number, related to driving thermal force), $\beta$ (aspect ratio of the convection layer). The simulator uses RK4 (dt = 0.005) to solve these equations numerically.

Real-World Applications

Weather & Climate Modeling: The Lorenz system was born from meteorology. It demonstrates the fundamental limits of long-term weather prediction due to chaos and is the starting point for thinking about the practical predictability horizon of the atmosphere.

Aerospace Engineering - Nonlinear Flutter: Nonlinear engineering models can show sensitivity, but they require their own equations, integration schemes and validation. This educational Lorenz model is not an aircraft design solver.

Structural Dynamics & Crash Simulation: In simulated car crash tests, the response can be highly sensitive to initial conditions (such as impact angle). Understanding chaos helps engineers account for variability and ensure robust safety designs.

Laser Physics & Electrical Circuits: Certain chaotic lasers and electronic oscillators are described by Lorenz-like equations. Engineers study these attractors to develop secure communication systems that use chaos to encrypt signals.

Common Misconceptions and Points to Note

First, understand that chaos does not equal randomness. The trajectory of the Lorenz attractor, while seemingly erratic, is generated completely deterministically from a fixed set of equations. If the initial values are the same, it will trace the exact same path every time. This concept of "deterministic chaos" is crucial in practical applications. For instance, if you run two CFD simulations under the same conditions and get slightly different results, you need the perspective to suspect that it might be due to slight mesh differences or numerical errors, rather than simply chalking it up to "unavoidable chaos."

Next, be aware of the "safe zones" and "danger zones" for parameter settings. The default values (σ=10, ρ=28, β=8/3) are the golden parameters where chaos is prominent. The Lorenz system is dissipative, so trajectories stay bounded even at large ρ; what needs care is that small changes in ρ can switch the behaviour (for σ=10, β=8/3, chaos and stable fixed points coexist for 24.06 < ρ < 24.74). In practical nonlinear analysis, too, it pays to locate the parameter ranges where the response changes character. The trick is to start with the default values, then change them incrementally (e.g., adjusting ρ in steps of 1 or 2) to observe the behavior.

Finally, understand that this simulator is a "visualization tool," not a "design tool." In practical CAE, you rarely solve the Lorenz equations themselves. However, the numerical integration methods like RK4 used here, and the "intuition" for understanding chaotic behavior, are extremely useful. For example, they build foundational skills for analyzing phenomena like how slight differences in initial conditions (such as passenger weight distribution) in a vehicle's nonlinear suspension vibration can lead to significant variation in the fatigue life of components after long-term driving.

How to Use

  1. Choose Classic chaos (σ=10, ρ=28, β=2.667) and compare two trajectories starting with Δx₀=10⁻⁵.
  2. In 3D, drag to rotate and pinch with two fingers to zoom. In 2D, a vertical swipe over the plot scrolls the page.
  3. Adjust trail length from 200 to 3000 points. Trail length and view remain adjustable while paused.
  4. Change ρ or σ one at a time and compare coordinates, separation Δ and λ. The finite-time estimate λ can fluctuate during short observations.

Worked Example

From the Classic chaos preset, change only σ to 4. With ρ=28 and β=2.667, the stable fixed points are C±=(±√(β(ρ−1)), ±√(β(ρ−1)), ρ−1)≈(±8.486, ±8.486, 27). Play long enough to see x and y approach about ±8.49 with the same sign and z approach 27. The selected side depends on the initial state. Restore σ=10 to compare stability at the same ρ.

Practical Notes

🎬 Watch it in motion

One Billionth Became 10,000 Different Futures
One Billionth Became 10,000 Different Futures
10 Satisfying Physics Simulations That Play Their Own Music
10 Satisfying Physics Simulations That Play Their Own Music
What Is the Lorenz Attractor? Chaos and the Butterfly Effect, Visualized
What Is the Lorenz Attractor? Chaos and the Butterfly Effect, Visualized
Chaos is NOT random. Change one billionth and watch. #Shorts
Chaos is NOT random. Change one billionth and watch. #Shorts
1,000x More Precision. Watch Which Pair Splits First. #Shorts
1,000x More Precision. Watch Which Pair Splits First. #Shorts
10,000 futures began within one billionth. Watch them split. #Shorts
10,000 futures began within one billionth. Watch them split. #Shorts
The shape is fixed. The position isn't. | Lorenz attractor #Shorts
The shape is fixed. The position isn't. | Lorenz attractor #Shorts
Not one random number inside | Lorenz attractor #Shorts
Not one random number inside | Lorenz attractor #Shorts