yet ALWAYS returns to start!
Period ≤ 3N iterations
Chaos that returns home — Vladimir Arnold's scrambling torus map
Vladimir Arnold illustrated this transformation in the 1960s using a picture of a cat. The map stretches and shears the image, then wraps it back onto the torus (mod N). After enough iterations, the scrambled mess magically reconstructs the original!
The map is genuinely chaotic: nearby points separate exponentially (Lyapunov exponent = ln(φ) where φ is the golden ratio!). But on a discrete N×N grid, there are only N² states—so it must eventually return to its starting point.
The matrix [[2,1],[1,1]] has determinant = 1, so area is preserved (it's a symplectic map). Points get scrambled but never created or destroyed. This connects to Hamiltonian mechanics and Liouville's theorem.
Send a scrambled image; the recipient applies more iterations until the original appears! The period depends on N (often much less than 3N), so knowing N lets you decrypt. Arnold's map inspired real encryption schemes.