HomeArticlesChaos & Dynamics

Strange Attractors: Order Hidden Inside Chaos

Lorenz, Rössler, Thomas and Halvorsen — how stretching and folding turn a deterministic system into a fractal trajectory that never repeats.

mysimulator teamUpdated June 2026≈ 8 min read▶ Open the simulation

Trajectories that never repeat and never escape

A strange attractor is what happens when a dynamical system is both dissipative (it loses energy or volume in phase space, so trajectories are pulled toward some smaller set) and chaotic (nearby trajectories diverge exponentially). The trajectory settles onto a bounded region it can never leave, yet inside that region it never closes into a repeating loop and never touches itself — the two requirements are only compatible if the attractor is a fractal, with a non-integer dimension, folded into itself infinitely many times.

live demo · a Lorenz-type trajectory folding through 3D space● LIVE

The Lorenz attractor: convection, reduced to three equations

Edward Lorenz derived his 1963 system as a drastic simplification of fluid convection in the atmosphere — three ordinary differential equations standing in for the temperature and velocity fields of a heated fluid layer:

dx/dt = σ(y - x)
dy/dt = x(ρ - z) - y
dz/dt = xy - βz

classic parameters:  σ = 10,  ρ = 28,  β = 8/3

For these parameters the trajectory spirals around one of two lobes, jumps unpredictably to the other, and never settles — the famous butterfly shape. Lorenz discovered sensitive dependence on initial conditions almost by accident, rerunning a simulation from rounded-off printed values and watching the new run diverge completely from the original within a simulated month. That observation is the origin of the popular "butterfly effect" phrase and the reason chaotic systems, however deterministic, are practically unpredictable beyond a finite horizon.

Rössler, Thomas and Halvorsen: different folds, same idea

The Rössler attractor (1976) was built deliberately to be the simplest possible chaotic flow with a single nonlinear term (a product xz), producing a single-lobe spiral that stretches and folds back on itself like a sheet of dough — easier to analyse geometrically than Lorenz's two-lobe system. René Thomas's cyclically symmetric attractor and the Halvorsen attractor use different nonlinear couplings between three variables but rely on the same recipe: stretching to separate nearby points exponentially (the source of chaos) and folding to keep the whole trajectory bounded (the source of the attractor's shape). Changing a single parameter — ρ in Lorenz, the coupling constants in Thomas — can move a system from a stable fixed point, through periodic orbits, into full chaos; the route usually passes through a cascade of period-doubling bifurcations.

Fractal dimension: more than a line, less than a surface

Because the attractor is stretched and folded infinitely, a cross-section through it looks like a Cantor-set of nested sheets rather than a simple curve. Its correlation dimension — a way of measuring how the number of nearby trajectory points scales with distance — comes out non-integer: about 2.06 for the classic Lorenz attractor. That fractional value is the mathematical fingerprint of "strange": an ordinary attractor (a point or a closed loop) has an integer dimension of 0 or 1, but a strange attractor sits between a surface and a solid, filling space more thoroughly than a sheet but never becoming a solid volume.

Reading the shape, not the numbers

Because two nearby chaotic trajectories diverge exponentially, tracking any single numerical value is pointless beyond a short horizon — the point of visualising these systems is the invariant shape the whole family of trajectories traces out, which stays the same however you perturb the starting point. Colour-by-time or colour-by-velocity rendering, and slow rotation, are what make the folding structure visible instead of just a tangled scribble.

Frequently asked questions

If the system is deterministic, why is it called chaotic?

Determinism only guarantees a unique future for an exact starting point. Because nearby starting points diverge exponentially, any tiny uncertainty in the initial condition (or floating-point rounding) grows until prediction becomes practically impossible, even though the equations themselves contain no randomness.

Why doesn't the trajectory ever repeat or cross itself?

The system is deterministic, so if the trajectory ever returned to an exact earlier state it would repeat forever from that point on — a periodic orbit, not chaos. A strange attractor avoids this by folding infinitely into a fractal structure, coming arbitrarily close to earlier points without ever landing on them exactly.

What decides whether the system settles into a fixed point, a cycle, or chaos?

The parameters of the equations, such as ρ in the Lorenz system. Small changes can push the system through bifurcations — from a stable point, to periodic oscillation, to full chaos — often via a period-doubling cascade that is a common route to chaos across many different physical systems.

Try it live

Everything above runs in your browser — open Strange Attractors and change the parameters while it is running. Nothing is installed, nothing is uploaded, the whole model lives in one tab.

▶ Open Strange Attractors simulation

What did you find?

Add reproduction steps (optional)