The maximum area of the convex hull of a polyhex
Determine the maximum area of the convex hull of a polyhex (an edge-connected set of n regular unit hexagons), proving a 2008 conjecture of Kurz.
The hull is written as a Minkowski sum P+H in axial coordinates on the triangular lattice, reducing area to widths of the cell-centre hull. Spanning-tree estimates bound P's area via edge counts, giving the tight upper bound M(n)=(1/6)ceil(n^2+14n/3). The upper bound was formalized in Lean 4 with Mathlib, defining cells as integer points, area as Lebesgue measure, and proving the hexagon has area one and the coordinate map has determinant one.
The maximum area equals (1/6)ceil(n^2+14n/3), confirming (and slightly sharpening) Kurz's conjecture, with a Lean development containing no unproved steps using only the three standard axioms. A brute-force enumeration over all polyhexes with at most 12 cells confirms the bound and shows a unique maximizing shape up to symmetry.
| n | fixed polyhexes | 6M(n) | maximisers (free) |
|---|---|---|---|
| 1 | 1 | 6 | 1 |
| 2 | 3 | 14 | 1 |
| 3 | 11 | 23 | 1 |
| 6 | 814 | 64 | 1 |
| 12 | 8182213 | 200 | 1 |
