The Hilbert Curve: How a Line Can Fill a Plane

Open with the geometric paradox: a line is 1D, a square is 2D, and dimension says one cannot equal the other. Pose the driving question and preview the recursive U-shape that will resolve it.

• A line has length but zero area; a square has area
• We are asking whether a single continuous path can pass through every point of a square
• The Hilbert curve is the canonical visual answer to this paradox

Ask the learner to commit to one intuition about the Hilbert curve before seeing any construction.

• Make an explicit prediction about what the curve looks like
• Test whether the learner assumes it is a spiral, a grid walk, or a true limit object

Let the learner choose the recursion order n and watch the Hilbert polyline unfold inside the unit square, with the path traced in real time and the grid cells visited highlighted.

• Each order divides the square into a 2^n by 2^n grid
• The path visits exactly one cell per step, traversing all 4^n cells without crossing itself
• Higher n visibly crowds more of the square interior

Display iterations n=1, 2, 3, and 4 next to each other so the learner can directly see how density grows while path length doubles each time.

• Order 1 is a simple U of three segments
• Each step turns every segment into four rotated copies
• Path length doubles (2^n) while bounding region stays fixed

Explain the substitution rule: take any Hilbert polyline, replace each of its 4^k segments with four quarter-sized U-copies rotated and flipped into the four quadrants, then rescale. This is the same rule at every scale, which is why the curve is a fractal.

• Self-similarity: the whole curve looks like four rotated copies of itself
• Continuity is preserved at every join by construction
• The limit curve is continuous because each iteration is continuous and they converge uniformly

Show that as n grows, the maximum distance from any point in the square to the nearest point on the curve shrinks toward zero, illustrating that the limit is genuinely space-filling.

• Pick any point in the square by clicking
• See the nearest segment of the Hilbert polyline light up
• Watch the distance drop sharply as n increases

Clarify common misconceptions: the limit is a curve, not a filled region; it is continuous but nowhere differentiable; it has a well-defined point for every t in [0,1], but it is not an injective map.

• The limit has the cardinality and topology of an interval, not of an area
• Multiple values of t can map to the same square point
• Space-filling here means the image covers the square, not that the curve has thickness

Transfer the idea: ask the learner to map a 16x16 grayscale image to a 1D Hilbert index and then 'unfold' it back, comparing locality preservation against a naive row-major scan.

• Hilbert order keeps spatially nearby pixels nearby in 1D
• Row-major order scatters neighbors far apart
• This is why Hilbert curves are used in spatial databases and image compression

Close the loop: restate the question, summarize that the Hilbert curve is a recursive fractal path whose limit is a continuous surjection from [0,1] onto the unit square, and note its real-world role in preserving locality.

• The Hilbert curve is a specific space-filling curve defined by a U-substitution rule
• Each iteration is a 1D path; the limit fills the 2D square
• It is a theoretical object with practical uses in mapping, indexing, and rendering