Sundog · Ghost
Ordered, but never repeating
A Penrose tiling is built from two rhombus shapes and a few local matching rules. It fills the plane with perfect order — yet no finite stamp ever repeats to cover it. Draw a circle anywhere below and watch what the circle can, and cannot, explain from the inside.
Click or drag on the tiling to move the circle. Raise the supertile level to see tiles gather into larger ancestors; widen the circle to pin down more of them.
Ordered does not mean repeating. Local does not mean closed.
The repeat cell that never comes
A wallpaper pattern closes its circle: capture one repeating cell and you have explained the whole wall. An aperiodic tiling keeps the circle open. Every finite patch is legal and concrete, but no finite patch is the whole law — the pattern is generated by a hierarchy of ever-larger supertiles rather than by a repeating unit. In the reader above, raising the supertile level shows that hierarchy: tiles merge into the larger ancestors they belong to.
But the outside is finite
"Open" is not "unbounded." Penrose tilings are recognizable: the supertile a patch belongs to is determined by a bounded neighbourhood around it. Widen the circle and the tiles inside stop crossing its edge — their place in the larger pattern becomes recoverable from the inside. You never have to look infinitely far. The context a patch is missing is a finite radius, and it recedes to a larger hierarchy as you zoom, rather than growing without bound.
How local is "bounded"? For the classic one-dimensional substitutions the radius is a couple of symbols; for this Penrose tiling it settles to about one tile-edge of neighbourhood. That number has a name in the literature — the constant of recognizability — and it is finite and computable.
What this is and is not
- It is an interactive reader for a known, published object (the rhombic Penrose tiling), generated from its substitution rule.
- It is not a new theorem, a new tiling, or a new invariant; the "bounded outside" is the known recognizability radius.
- Aperiodic is not random: the tiling is perfectly ordered, just not periodic.
- The picture is an intuition surface, not a proof, and the page makes no claim about Goedel, incompleteness, or self-reference.
Sources & further reading
- Penrose rhombus tiling: R. Penrose (1974); M. Gardner, Scientific American (1977). Geometry here is generated from the published substitution rule; no figures are reproduced.
- Recognizability / constant of recognizability: B. Mosse; F. Durand & J. Leroy, "The constant of recognizability is computable for primitive morphisms" (arXiv:1610.05577).
- Sundog write-up: the Ghost ledger and the recognizability-radius memo.
A Sundog reader/exhibit. Names a known boundary; claims no new mathematics.