Fractal

6 min read#chaos

A shape that repeats its structure at every scale, with a fractional dimension that measures how intricately it fills space.

Contents

Fractal

A fractal is a shape that looks the same — or statistically similar — no matter how far you zoom in. Coastlines, ferns, lungs, lightning, and the cross-section of every Strange Attractor share this property of self-similarity: the whole is echoed in its parts, down to arbitrarily small scales. Fractals are the natural geometry of chaos, because the stretch-and-fold dynamics that generate chaos lay down structure at every scale at once.

Self-similarity and infinite detail

Zoom into a smooth curve and it eventually looks like a straight line — it is locally simple. Zoom into a fractal and the complexity never resolves; new detail keeps appearing. The classic constructions make this concrete: the Koch curve adds a bump to every segment forever, becoming infinitely long while enclosing a finite area; the Cantor set repeatedly deletes middle thirds, leaving infinitely many points of total length zero.

The Julia set below is the showpiece. Each point of the plane is colored by how quickly the iteration z \mapsto z^2 + c runs away to infinity; the black region is the set of points that never escape. Its boundary is an infinitely detailed fractal whose shape is dictated entirely by the constant c.

Click and drag on the image to change c and watch the fractal morph — from connected filaments to exploding dust — in real time.

Filled Julia set of z → z² + c. Drag anywhere to steer c by mouse position; release to keep it. The black interior never escapes; the glowing boundary is fractal at every scale.

Fractal dimension

How do you measure something rougher than a curve but thinner than a region? With a dimension that need not be a whole number. The box-counting dimension asks how the number N(\varepsilon) of little boxes of size \varepsilon needed to cover the set grows as the boxes shrink:

D = \lim_{\varepsilon \to 0}\frac{\ln N(\varepsilon)}{\ln(1/\varepsilon)}

For a smooth line you get D = 1, for a filled square D = 2. The Koch curve gives D = \ln 4/\ln 3 \approx 1.26 — genuinely between a line and an area. The cross-section of the Lorenz attractor sits at D \approx 2.06. A fractional dimension is precisely the quantitative statement that detail persists across scales.

The Koch curve has box-counting dimension about 1.26. What does a non-integer dimension between 1 and 2 tell you?

See also