Strange Attractors
Chaotic trajectories that never repeat but never escape — drag to rotate in 3D, tune parameters to deform the attractor
Morph A 1.00
Morph B 1.00
Points 200,000
Lorenz
Attractor
0
Points
0
FPS
Lorenz Attractor — Edward Lorenz (1963), modelling atmospheric convection: ẋ = σ(y−x), ẏ = x(ρ−z)−y, ż = xy−βz. With σ=10, ρ=28, β=8/3 the trajectory never repeats yet stays bounded — the first mathematical proof of deterministic chaos. Drag to orbit, scroll to zoom.

About this simulation

This page integrates five classic chaotic systems — Lorenz, Aizawa, Halvorsen, Thomas, Dadras — with a 4th-order Runge-Kutta solver and plots up to 300,000 visited points as an additive-blended 3D point cloud in WebGL. Each system is a set of three coupled nonlinear ODEs: tiny changes in the starting point fan out into wildly different trajectories, yet the long-run path stays confined to a bounded, self-similar fractal shape — the strange attractor itself.

🔬 What it shows

Deterministic chaos: fully predictable equations (no randomness anywhere) that still produce trajectories which never repeat and are practically impossible to forecast far ahead, because nearby paths diverge exponentially.

🎮 How to use

Pick a preset (Lorenz, Aizawa, Halvorsen, Thomas, Dadras), then drag the Morph A / Morph B sliders to rescale key parameters like σ, ρ or b and watch the shape deform live. The Points slider trades render density for speed, Autorotate spins the camera, and Reset re-integrates from the initial condition.

💡 Did you know?

The Lorenz attractor came from a 1963 weather-modelling accident: Edward Lorenz re-entered a rounded mid-run number and got a completely different forecast, coining what we now call the butterfly effect.

Frequently asked questions

What is a strange attractor?

It's the bounded, fractal-shaped region of space that a chaotic dynamical system's trajectory settles onto and wanders through forever without ever exactly repeating or leaving the region.

Why do the trajectories never intersect themselves?

For a deterministic system, one state determines exactly one future path — if two trajectory segments crossed at the same point, they would have to follow identically from then on, so in continuous chaotic flows distinct trajectories never truly cross.

What do the Morph A and Morph B sliders actually change?

They scale a real coefficient in that attractor's equations — for Lorenz these are the σ (Prandtl-like) and ρ (Rayleigh-like) constants; changing them can shift the system between chaotic, periodic, or divergent regimes.

Why does increasing the Points slider slow things down?

Each point requires one RK4 integration step (four equation evaluations) plus a GPU vertex upload, so render and compute cost scale roughly linearly with the point count, up to 300,000.

Is this simulating real physics?

Yes for Lorenz, which models simplified atmospheric convection; the others (Aizawa, Halvorsen, Thomas, Dadras) are mathematically constructed chaotic systems studied for their geometric and dynamical properties rather than a specific physical process.