In 1963, Edward Lorenz discovered that three simple equations — no randomness, no noise — could produce behavior that was fundamentally unpredictable. The rules are fixed. The outcome is chaos.
scroll to begin
In 1963, meteorologist Edward Lorenz built a simplified model of atmospheric convection. Three variables — proportional to convection intensity, horizontal temperature variation, and vertical temperature variation — linked by three differential equations.
The system is entirely deterministic. Given a starting point, the future is mathematically fixed. There is no randomness anywhere.
The Lorenz system: dx/dt = σ(y − x) dy/dt = x(ρ − z) − y dz/dt = xy − βz
Three variables, three parameters (σ = 10, ρ = 28, β = 8/3), and nothing else. Pick any starting point and the equations dictate exactly where it goes.
Start at a single point and let the equations evolve it forward. The cyan trail traces the path through three-dimensional state space.
Watch it spiral outward on one wing, then suddenly jump to the other. Back and forth, unpredictably.
The trajectory never settles to a fixed point. It never repeats. Instead, it traces an intricate shape — two lobes like the wings of a butterfly — orbiting one wing, then the other, switching unpredictably.
This shape is the Lorenz attractor. It is "strange" in a precise sense: it has fractal dimension, approximately 2.06.
Now start two trajectories from nearly identical points — cyan and orange — separated by just 0.001 in one coordinate.
At first, they trace the same path. They appear to be the same trajectory.
Wait. Within a few orbits, the two paths begin to separate. One loops left while the other loops right. Within seconds they are on completely different parts of the attractor.
The distance between them grows exponentially — doubling roughly every 0.78 time units.
This is sensitive dependence on initial conditions — what Lorenz called "the butterfly effect." An immeasurably small change in starting conditions leads to a completely different future.
The equations haven't changed. The rules are the same. But the prediction is useless.
Long-range weather prediction is impossible — not because we don't know the equations, but because we can never measure the initial conditions precisely enough. Determinism does not guarantee predictability.
The boundary between order and chaos is a property of the system itself.
In 1961, Lorenz was re-running a weather simulation when he entered initial conditions rounded to three decimal places instead of six. He expected the same output. What he got was a completely different weather pattern.
That moment — noticing that 0.506 and 0.506127 produced entirely different futures — launched the modern study of chaos. The sensitivity was not a bug. It was a property of the equations themselves.
Each line tracks the x-coordinate of a trajectory starting from nearly the same initial point. They track together, then fan out completely.
25 trajectories. Initial conditions within 10⁻⁴ of (1, 1, 1). σ = 10, ρ = 28, β = 8/3.
The Lorenz attractor occupies zero volume — the trajectory is confined to a set of measure zero in three-dimensional space. Yet it never repeats. The path loops endlessly through the two wings, tracing an infinitely long curve on an infinitely thin folded surface.
This combination — bounded but aperiodic — is the hallmark of deterministic chaos. The system is constrained, but within those constraints, it is unpredictable forever.
Plot each successive maximum of z against the one before it. Despite the chaos, the points fall on a nearly one-dimensional curve — the Lorenz map.
5,000 successive z-maxima from a single trajectory. The near-1D structure means the chaos is governed by a simple stretching-and-folding mechanism.
Lorenz published "Deterministic Nonperiodic Flow" in the Journal of the Atmospheric Sciences in 1963. The paper took over a decade to find its audience, but eventually became one of the most cited works in mathematics and physics.
The Lorenz system is now a canonical example in dynamical systems theory, and its "butterfly effect" metaphor has entered everyday language. Variants of the model appear in fluid dynamics, laser physics, chemical reactions, and economic modeling.
Adjust the parameters and watch the attractor reshape. Push ρ below 24.7 and the chaos disappears — the system settles to a fixed point. Push it higher and the butterfly stretches.