Aperiodic order from periodic higher dimensions. The Penrose tiling as a shadow of a 5D lattice.
On April 8, 1982, Dan Shechtman saw a diffraction pattern with ten-fold rotational symmetry from a rapidly cooled Al-Mn alloy. Classical crystallography forbids five-, eight-, ten-, and twelve-fold rotational symmetries in periodic structures. What Shechtman saw was a quasicrystal — long-range order without translational periodicity.
The result was attacked for years. Linus Pauling famously declared "there are no quasicrystals, only quasi-scientists." The IUCr eventually rewrote its definition of a crystal. Shechtman received the 2011 Nobel Prize in Chemistry.
The mathematical mechanism is cut-and-project: a quasicrystal is the shadow of a higher-dimensional lattice sliced by an irrationally angled hyperplane. Aperiodicity comes from the irrational angle, not from disorder.
This maps directly onto parallel coordinates brushing: the "acceptance window" that selects which lattice points project is a range constraint on the perpendicular-space axes — the same operation as axis brushing.
Series note. This chapter is the parallel-coordinates doorway into the topic. The dedicated Aperiodic Order: Quasicrystals from Lattice Projections series rebuilds the story across four articles and tightens the labels around phasons, diffraction, Penrose projection, and the icosian / nearby E8 substrate.
The simplest quasicrystal lives in one dimension. Start with the integer lattice in the plane. Tilt a strip at an irrational angle, specifically at slope , where is the golden ratio. Every lattice point inside the strip gets projected down onto the strip's long axis. The result is a sequence of two interval lengths: Long and Short. This sequence is the Fibonacci chain.
The Fibonacci chain never repeats but isn't random. The substitution L → LS, S → L generates it. The L-to-S ratio approaches φ as the chain grows.
Figure 1. Left: the lattice with a strip at slope . Lattice points inside the strip (blue) project onto the physical axis to produce the Fibonacci chain. Right: 2-axis parallel coordinates of the same lattice points. The acceptance window is the range constraint on the perpendicular axis. This is brushing. Adjust the strip width to change the window.
The vertical range on the perpendicular axis is the acceptance window. Widening admits more lattice points; narrowing removes them. This is axis brushing applied to a lattice.
The Fibonacci chain lives in 1D. To build a 2D quasicrystal, we need a higher-dimensional lattice. The Penrose tiling, the most celebrated quasicrystal, is a projection from , the integer lattice in five dimensions.
The projection uses the fifth roots of unity. For each basis vector in , we define two projection directions:
The first pair gives the "physical space" coordinates; the second pair gives the "perpendicular space" coordinates. A lattice point is accepted if its perpendicular projection falls inside the acceptance window, a decagonal region in the 3D perpendicular space (though we typically check a 2D cross-section). The accepted points project onto the physical plane, where they form the vertices of two types of rhombi: thick (with angles 72° and 108°) and thin (with angles 36° and 144°).
Figure 2. Penrose tiling generated by the cut-and-project method from . Thick rhombi (blue) and thin rhombi (gold). Adjust the acceptance window radius and patch size.
Figure 3. 5-axis parallel coordinates of the lattice points whose perpendicular projections fall inside the acceptance window. Each polyline is a 5D integer vector that maps to a vertex in the Penrose tiling above.
The acceptance window is the heart of the cut-and-project method. It is a region in the perpendicular space. Only lattice points whose perpendicular projection lands inside the window survive to be projected into the physical tiling.
For the Fibonacci chain, the window is a simple interval on the perpendicular axis. Shifting the window along the perpendicular direction changes which lattice points are selected. The tiling rearranges locally (some points enter, others leave) but the global statistical properties remain the same. This is called a phason shift.
Figure 4. Left: the acceptance window (blue band) shifts along the perpendicular axis. Points enter and exit. Right: the resulting Fibonacci chain updates in real time. The sequence changes locally, but the ratio of L to S intervals stays near φ. This continuous shift is a phason.
Quasicrystals produce sharp Bragg peaks in their diffraction patterns, just like periodic crystals. But the peaks are arranged with non-crystallographic symmetry. The Penrose tiling's diffraction has ten-fold rotational symmetry: the pattern that Shechtman saw in 1982.
The diffraction pattern is the squared modulus of the Fourier transform of the point set. For a periodic crystal, the peaks form a regular reciprocal lattice. For a quasicrystal, the peaks are dense but discrete, indexed by integer linear combinations of incommensurate basis vectors. For a random point set, there are no sharp peaks at all, just a diffuse ring.
Figure 5. Diffraction patterns. The quasicrystal (Penrose vertices) produces sharp peaks with 10-fold symmetry. A periodic crystal produces a regular grid of peaks. A random point set produces a diffuse ring. Toggle between modes.
The golden ratio saturates every level of the Penrose construction. The projection slope for the Fibonacci chain is . The ratio of the long interval to the short interval is . The areas of the thick and thin rhombi have ratio . And the ratio of thick to thin tiles in an infinite Penrose tiling is exactly .
The golden ratio is the eigenvalue of the projection. The matrix splitting 5D into physical and perpendicular subspaces has φ and 1−φ as eigenvalues. The Fibonacci recurrence reflects the self-similar substitution that generates the tiling.
Figure 6. As the Penrose patch grows, the ratio of thick to thin rhombi converges to φ. The dashed line marks the golden ratio. Each step adds a ring of tiles and the running ratio oscillates around φ, approaching it from alternating sides, exactly as Fibonacci ratios do.
The icosian / Elser-Sloane thread sits close to E8's root system, and the parallel-coordinates sketch below uses E8 roots as a useful visual proxy. The fuller quasicrystal series uses a stricter label: the in-house model-set substrate is Hurwitz-golden and Galois-pair first, with E8 nearby and related rather than the canonical source being projected.
The result is related to the 600-cell (120 vertices, 600 tetrahedral cells, icosahedral symmetry). The two 4D images are related by φ. The Elser–Sloane quasicrystal connects E8's exceptional geometry to the five-fold symmetry of quasicrystals through the same cut-and-project mechanism.
Figure 7. Left: 8-axis parallel coordinates of the 240 E8 root vectors. Right: after the cut-and-project split into 4 physical + 4 perpendicular axes, the surviving roots shown in 4-axis PC. The acceptance window in perpendicular 4D selects a subset with icosahedral symmetry.
Real icosahedral quasicrystals (Shechtman's Al-Mn alloy and hundreds of compounds since) live in 3D. The bridge from 4D to 3D is another irrational projection. Projecting the 4D quasicrystal along the direction whose cosine with the projection axis is 1/(φ+2) yields the Fibonacci icosagrid: a 3D aperiodic point set with full icosahedral symmetry.
The icosagrid's nucleus is the Compound of Five Cuboctahedra: 5 cuboctahedra (12 vertices each) arranged around a common center by the icosahedron's five-fold rotations. The 60 vertices form a rigid scaffold with symmetry group Ih (order 120).
Figure 8. The Compound of Five Cuboctahedra (C5C), atomic nucleus of the 3D Fibonacci icosagrid. Each cuboctahedron (12 vertices, 24 edges) is a copy of a base figure rotated by 2π/5 around one of the icosahedron's six five-fold axes. The five copies share a common centre but no other vertex; the union has 60 vertices and the icosahedral point group Ih. Cycle through the cuboctahedra to watch them interlock.
The chain in this article is a visual bridge: E8-like 8D data -> an Elser-Sloane-style 4D view -> a 3D icosahedral sketch -> C5C as local nucleus. The dedicated series separates that bridge from the stricter arithmetic model so the reader can see what is proven, what is a related realization, and what is a useful visual analogy.
The Ammann-Beenker tiling has eight-fold symmetry, projecting from the 4D integer lattice using eighth roots of unity. Tiles are a square and a 45° rhombus, governed by the silver ratio (instead of φ).
Figure 8. Left: Ammann-Beenker tiling from the projection. Squares (green) and 45° rhombi (purple). Right: 4-axis parallel coordinates of the lattice points. Compare with the 5-axis PC of the Penrose: the structure is analogous but the symmetry is octagonal rather than pentagonal.
The pattern. Fibonacci chain: projected to 1D via the golden ratio. Penrose tiling: projected to 2D via fifth roots of unity. Ammann-Beenker: projected to 2D via eighth roots. Elser-Sloane: E8 projected to 4D. In each case, the construction is the same: a higher-dimensional periodic lattice, an irrational projection, and an acceptance window. And in each case, the acceptance window is a brushing operation on the perpendicular-space coordinates in parallel coordinates.
McKay correspondence. The five-fold symmetry is not arbitrary. Finite subgroups of SU(2) — cyclic, binary dihedral, binary tetrahedral, binary octahedral, binary icosahedral — are in bijection with the affine simply-laced Dynkin diagrams Ãn, &Dtilde;n, &Etilde;6, &Etilde;7, &Etilde;8. The binary icosahedral group 2I (120 elements, the unit icosians, the vertices of a 600-cell) corresponds to affine &Etilde;8. So the same 120-element group that sits at the core of the Elser-Sloane projection also generates E8's root system through the Borel–de Siebenthal machine: 9-node affine Dynkin → prune a node → sub-rootsystem. Icosahedral quasicrystal order and E8's arithmetic are two views of the same finite group.
The quasicrystal story connects to several threads in this series. The brushing operation that defines the acceptance window. The E8 root system whose projection yields icosahedral quasicrystals. The polytopes that appear as cross-sections of the projected structures. And looking ahead, error-correcting codes are lattices too, and the same projection machinery connects coding theory and quasicrystal physics.
What Shechtman saw in 1982 was a shadow. Not a shadow of disorder, but a shadow of higher-dimensional order, perfect periodicity in a space we cannot directly see, projected into a pattern that refuses to repeat.