A periodic twelve-fold tiling by the Spectre
The Spectre of Smith, Myers, Kaplan and Goodman-Strauss tiles the plane by rotations and translations, and every such tiling is non-periodic. Andrew Bayly asked on MathOverflow (question 516016) whether it admits a periodic n-fold tiling for some n > 1, in which almost every point lies in exactly n tiles. It does, with n = 12: we give 54 rotated and translated copies of the Spectre whose translates by a lattice cover almost every point exactly 12 times, with no reflections (for the straight-edged Spectre no two tiles coincide). The proof is a finite check: the 756 edges of one period cell cancel in oppositely oriented pairs modulo the lattice, so the covering number is constant, and comparing areas gives 12. The construction works for the straight-edged Tile(1,1) and for every curved Spectre of the common-curve family. Whether some 1 < n < 12 is possible is open. The construction was first announced in the author's MathOverflow answer 516104 (16 October 2026). For the straight-edged Spectre the theorem is formally verified in Lean 4 (standard axioms only); for curved Spectres the edge-record certificate is kernel-checked, and the step to constant coverage, proved in the paper by Green's theorem, is not formalised. The archive holds the LaTeX source, the Lean development with mutation tests, an exact-arithmetic checker and an independent Monte Carlo count (code under the MIT licence). Version 2 of the preprint (version 1: 20 October 2026); identical to the folder spectre-twelve-fold at tag spectre-twelve-fold-v2 in github.com/jvvk/mathematics.
Authors
- Vamshi Jandhyala (ORCID: https://orcid.org/0009-0006-6708-5805)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-21
- DOI
- https://doi.org/10.5281/zenodo.24406774
- Primary Topic
- Quasicrystal Structures and Properties
- Type
- preprint