Aperiodic order is a crystal seen from a higher dimension

A Penrose tiling looks paradoxical: it is perfectly ordered — every tile placed by a strict rule — yet it never repeats. Where does that ordered-but-non-repeating structure come from? The most beautiful answer in the whole subject is this: an aperiodic pattern in low dimensions is the shadow of a perfectly periodic crystal living in a higher dimension. Tilt an ordinary repeating lattice, slice a thin slab of it, cast that slab's points down onto a lower- dimensional screen — and out falls a pattern that is aperiodic, yet as orderly as the crystal it came from.

This recipe is called the cut-and-project method, and the patterns it produces are called model sets. It does two jobs at once. It explains where the hidden order in a quasicrystal comes from — it is simply inherited from a genuine periodic lattice one dimension up. And it gives us a machine for building such patterns to order: choose the lattice, choose the slice, turn the handle.

The recipe: cut, then project

Every cut-and-project construction follows the same four moves. Two of them set the stage; the last two — the cut and the project — give the method its name.

• 1. Start with a periodic lattice. Take the integer lattice in some higher-dimensional space — the perfectly regular grid of points with whole-number coordinates, \mathbb{Z}^n. Nothing exotic: it is the most boring, most repetitive object there is.
• 2. Choose a subspace — the "physical" space — at an irrational slope. Pick a lower-dimensional flat (a line, or a plane) through the lattice, and orient it so its slope relative to the lattice is irrational. This single choice is where all the magic hides, as we will see.
• 3. Cut. Keep only the lattice points that lie inside a thin strip hugging that subspace — a band of fixed thickness called the acceptance window. Every other lattice point is discarded.
• 4. Project. Drop each surviving point straight onto the subspace. The resulting cloud of points on the subspace is the model set — an aperiodic, highly ordered pattern.

Because the surviving points came from a perfect lattice, the result is exquisitely ordered. Because the slope is irrational, the strip never lines up the same way twice, so the projected pattern never repeats. Order from the lattice, aperiodicity from the irrational slope — the two ingredients that seemed to contradict each other, delivered together by one construction. The parent space you need is always just a couple of dimensions higher than the pattern you want.

Worked example 1 — the Fibonacci chain from a 2-D lattice

The cleanest case lives one dimension up from a line. We will build the Fibonacci chain: an infinite row of two kinds of interval — a Long one (L) and a Short one (S) — laid end to end in the never-repeating order

Here is the construction, step by step.

• The lattice is the ordinary square grid of integer points (m, n) in the plane — a two-dimensional periodic crystal.
• The physical space is a single line through the origin whose slope is 1/\varphi \approx 0.618, where \varphi = \tfrac{1+\sqrt{5}}{2} is the golden ratio. Since \varphi is irrational, this line has irrational slope: it passes through the origin but never hits another lattice point again, ever.
• Cut: lay a strip of one lattice-cell's width along that line, and keep only the grid points falling inside it. Those survivors form a staircase that climbs alongside the line, taking a rightward step here, an upward step there.
• Project: drop each staircase point onto the line. A rightward grid-step projects to a slightly longer gap; an upward grid-step projects to a slightly shorter gap. Read the gaps off in order and you get exactly L\,S\,L\,L\,S\,L\,S\ldots — the Fibonacci chain.

The two lengths are just the shadows of the grid's two unit steps, so their ratio is fixed forever at the golden ratio itself:

Because that ratio is irrational, no block of Ls and Ss can ever tile the line by simple repetition — precisely the same golden-ratio argument that makes the Penrose tiling aperiodic, now playing out on a line. Step through the figure below to watch all four moves happen.

Watch it built

The faint dots are the integer grid. Play the figure: the slanted line is the physical space at slope 1/\varphi; the band is the acceptance window; the highlighted dots are the points the cut keeps; the coloured staircase edges are their rightward (L) and upward (S) steps; and the final step drops everything onto the line to reveal the L/S chain.

Notice that the chain of Ls and Ss down the line never settles into a repeating block — yet every single interval was dictated by the rigid grid above it. That is aperiodic order in one picture.

Why the slope must be irrational

The irrational slope is doing all the work, and it is worth seeing exactly why. Suppose we had cheated and used a rational slope instead — say a line of slope 2/3. A line of slope 2/3 through the origin passes through the lattice point (3, 2), and then (6, 4), and then (9, 6) — it hits a grid point every three steps across, on the dot, forever. The strip around it therefore repeats with period (3, 2), and the projected pattern is a plain periodic row: the same finite block of gaps, over and over. Ordinary, boring, repeating.

An irrational slope like 1/\varphi can never do that. After the origin, the line misses every lattice point for the rest of time — there is no (p, q) it ever lands on again, because that would force 1/\varphi = q/p, a fraction, contradicting irrationality. With no exact alignment ever recurring, the strip's contents never repeat, and the projected chain is aperiodic. Irrational in, aperiodic out; rational in, periodic out. That is the entire secret of the method in one sentence.

Worked example 2 — Penrose is a projection from five dimensions

Now step up a dimension in the output. To get an aperiodic tiling of the plane with five-fold symmetry, we run the very same recipe from a much roomier parent: the integer lattice in five dimensions, \mathbb{Z}^5. A Penrose tiling is the cut-and-project image of the 5-D cubic lattice down onto a cleverly chosen plane.

Why exactly five? Because the five-fold symmetry of the tiling is inherited from the symmetry of the 5-D cube. The five coordinate axes of \mathbb{Z}^5 can be cycled into one another, and when you project onto the right plane those five axes come down as five directions 72^\circ apart — the spokes of a five-pointed star. The "forbidden" symmetry the plane could never support on its own is perfectly natural upstairs, and the projection simply carries it down.

And the acceptance window earns its keep here. In the Fibonacci case the window was just an interval, and its width set the two gap lengths. For Penrose the window is a little two-dimensional region (a pentagon-shaped patch), and its shape decides which tiles appear and how they fit — change the window and you change the tile set. The window is the dial that tunes exactly which model set you build.

Why quasicrystals give sharp diffraction

This viewpoint quietly solves a real-world puzzle. When Dan Shechtman fired electrons at a quasicrystal, he saw a sharp, spotty diffraction pattern — the crisp bright dots that everyone believed only a periodic crystal could produce. How can a pattern that never repeats scatter waves into clean, sharp spots?

Cut-and-project answers it in a line: the sharpness is inherited from the parent lattice. A perfectly periodic lattice diffracts into sharp spots — that is textbook reciprocal-lattice physics. Projecting the lattice down to build the quasicrystal carries that sharp spottiness down with it (the spots simply become denser and land at golden-ratio-spaced positions). So the quasicrystal diffracts sharply for the same reason the Fibonacci chain is ordered: it is a shadow of something perfectly periodic one dimension up. Sharp diffraction is the fingerprint of that hidden higher-dimensional order.