A topic in the Open Knowledge Graph — a free, open map of 15,290 topics and the order to learn them in.
A strange attractor is a bounded set in phase space that attracts nearby trajectories, has a fractal (non-integer) dimension, and exhibits sensitive dependence on initial conditions. It is "strange" because of its fractal geometry — it has zero volume but infinite detail, with self-similar structure at all scales. Strange attractors are the geometric signature of dissipative chaos: the stretching that produces sensitivity and the folding that maintains boundedness create an infinitely layered, Cantor-set-like structure.
An attractor is a set in phase space where trajectories end up — the long-term fate of the dynamics. You've encountered simple attractors: stable fixed points (zero-dimensional), stable limit cycles (one-dimensional closed curves). Strange attractors are the chaotic analog: sets that attract trajectories but have a complicated, fractal internal structure that defies description as a smooth manifold. They are the geometric objects on which chaotic dynamics lives.
The "strangeness" is geometric: the attractor has a fractal (non-integer) dimension. The Lorenz attractor, with dimension approximately 2.06, is slightly more than a surface but vastly less than a volume. If you zoom into any piece of it, you find infinitely many nearly-parallel sheets separated by gaps, like a book with infinitely many infinitely-thin pages. This structure arises from the stretching-and-folding mechanism of chaos. Each pass through the attractor stretches a piece of the trajectory in one direction and compresses it in another, then folds it back. Repeated stretching and folding produces a layered structure at every scale — a Cantor-set cross-section in the direction of compression, smooth in the direction of stretching.
The fractal dimension captures this structure quantitatively. For the Lorenz system, the Kaplan-Yorke dimension D_KY ≈ 2 + 0.9/14.6 ≈ 2.06 reflects the competition between stretching (λ₁ ≈ +0.9) and compression (λ₃ ≈ -14.6). The stretching slightly overpowers what a surface would need, inflating the dimension just above 2. The compression is so strong that the attractor remains very close to two-dimensional — it's almost a surface, but not quite. Different strange attractors have different fractal dimensions depending on their Lyapunov spectrum: the Rossler attractor is about 2.01 (barely strange), while the hyperchaotic attractors found in higher-dimensional systems can have dimensions of 3 or more.
Strange attractors resolve a paradox: how can a dissipative system (which contracts volumes) have persistent, complex dynamics? If volumes shrink, shouldn't everything collapse to a point? The resolution is that volume contraction and dimension are different things. Volumes shrink to zero, but the surviving set — the attractor — can have complex geometry. Think of repeatedly stretching and folding a piece of dough: its volume stays roughly constant (for incompressible dough), but its structure becomes infinitely complex. For dissipative systems, the dough also gets thinner with each fold, so the volume shrinks to zero — but the layered, fractal structure remains. The strange attractor is what's left when you've squeezed out all the volume but kept all the complexity.